jlengrand/MachineLearningCoursera
1
0
Fork 0
mirror of https://github.com/jlengrand/MachineLearningCoursera.git synced 2026-03-10 08:31:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

Starting to work on Week 7.

Browse Source 
Yeah i know i'm super late.
Those weeks go way to fast.
This commit is contained in:
julien Lengrand-Lambert
2015-12-19 18:25:05 +01:00
parent cfcd8b02b0
commit 8c6997b185
41 changed files with 6192 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

BIN
machine-learning-ex6.zip Normal file
View File

Binary file not shown.

BIN
machine-learning-ex6/ex6.pdf Normal file
View File

Binary file not shown.

34
machine-learning-ex6/ex6/dataset3Params.m Normal file
View File

@@ -0,0 +1,34 @@
function [C, sigma] = dataset3Params(X, y, Xval, yval)
%EX6PARAMS returns your choice of C and sigma for Part 3 of the exercise
%where you select the optimal (C, sigma) learning parameters to use for SVM
%with RBF kernel
% [C, sigma] = EX6PARAMS(X, y, Xval, yval) returns your choice of C and
% sigma. You should complete this function to return the optimal C and
% sigma based on a cross-validation set.
%
% You need to return the following variables correctly.
C = 1;
sigma = 0.3;
% ====================== YOUR CODE HERE ======================
% Instructions: Fill in this function to return the optimal C and sigma
% learning parameters found using the cross validation set.
% You can use svmPredict to predict the labels on the cross
% validation set. For example,
% predictions = svmPredict(model, Xval);
% will return the predictions on the cross validation set.
%
% Note: You can compute the prediction error using
% mean(double(predictions ~= yval))
%
% =========================================================================
end

61
machine-learning-ex6/ex6/emailFeatures.m Normal file
View File

@@ -0,0 +1,61 @@
function x = emailFeatures(word_indices)
%EMAILFEATURES takes in a word_indices vector and produces a feature vector
%from the word indices
% x = EMAILFEATURES(word_indices) takes in a word_indices vector and
% produces a feature vector from the word indices.
% Total number of words in the dictionary
n = 1899;
% You need to return the following variables correctly.
x = zeros(n, 1);
% ====================== YOUR CODE HERE ======================
% Instructions: Fill in this function to return a feature vector for the
% given email (word_indices). To help make it easier to
% process the emails, we have have already pre-processed each
% email and converted each word in the email into an index in
% a fixed dictionary (of 1899 words). The variable
% word_indices contains the list of indices of the words
% which occur in one email.
%
% Concretely, if an email has the text:
%
% The quick brown fox jumped over the lazy dog.
%
% Then, the word_indices vector for this text might look
% like:
%
% 60 100 33 44 10 53 60 58 5
%
% where, we have mapped each word onto a number, for example:
%
% the -- 60
% quick -- 100
% ...
%
% (note: the above numbers are just an example and are not the
% actual mappings).
%
% Your task is take one such word_indices vector and construct
% a binary feature vector that indicates whether a particular
% word occurs in the email. That is, x(i) = 1 when word i
% is present in the email. Concretely, if the word 'the' (say,
% index 60) appears in the email, then x(60) = 1. The feature
% vector should look like:
%
% x = [ 0 0 0 0 1 0 0 0 ... 0 0 0 0 1 ... 0 0 0 1 0 ..];
%
%
% =========================================================================
end

10
machine-learning-ex6/ex6/emailSample1.txt Normal file
View File

@@ -0,0 +1,10 @@
> Anyone knows how much it costs to host a web portal ?
>
Well, it depends on how many visitors you're expecting.
This can be anywhere from less than 10 bucks a month to a couple of $100.
You should checkout http://www.rackspace.com/ or perhaps Amazon EC2
if youre running something big..
To unsubscribe yourself from this mailing list, send an email to:
groupname-unsubscribe@egroups.com

34
machine-learning-ex6/ex6/emailSample2.txt Normal file
View File

@@ -0,0 +1,34 @@
Folks,
my first time posting - have a bit of Unix experience, but am new to Linux.
Just got a new PC at home - Dell box with Windows XP. Added a second hard disk
for Linux. Partitioned the disk and have installed Suse 7.2 from CD, which went
fine except it didn't pick up my monitor.
I have a Dell branded E151FPp 15" LCD flat panel monitor and a nVidia GeForce4
Ti4200 video card, both of which are probably too new to feature in Suse's default
set. I downloaded a driver from the nVidia website and installed it using RPM.
Then I ran Sax2 (as was recommended in some postings I found on the net), but
it still doesn't feature my video card in the available list. What next?
Another problem. I have a Dell branded keyboard and if I hit Caps-Lock twice,
the whole machine crashes (in Linux, not Windows) - even the on/off switch is
inactive, leaving me to reach for the power cable instead.
If anyone can help me in any way with these probs., I'd be really grateful -
I've searched the 'net but have run out of ideas.
Or should I be going for a different version of Linux such as RedHat? Opinions
welcome.
Thanks a lot,
Peter
--
Irish Linux Users' Group: ilug@linux.ie
http://www.linux.ie/mailman/listinfo/ilug for (un)subscription information.
List maintainer: listmaster@linux.ie

150
machine-learning-ex6/ex6/ex6.m Normal file
View File

@@ -0,0 +1,150 @@
%% Machine Learning Online Class
% Exercise 6 | Support Vector Machines
%
% Instructions
% ------------
%
% This file contains code that helps you get started on the
% exercise. You will need to complete the following functions:
%
% gaussianKernel.m
% dataset3Params.m
% processEmail.m
% emailFeatures.m
%
% For this exercise, you will not need to change any code in this file,
% or any other files other than those mentioned above.
%
%% Initialization
clear ; close all; clc
%% =============== Part 1: Loading and Visualizing Data ================
% We start the exercise by first loading and visualizing the dataset.
% The following code will load the dataset into your environment and plot
% the data.
%
fprintf('Loading and Visualizing Data ...\n')
% Load from ex6data1:
% You will have X, y in your environment
load('ex6data1.mat');
% Plot training data
plotData(X, y);
fprintf('Program paused. Press enter to continue.\n');
pause;
%% ==================== Part 2: Training Linear SVM ====================
% The following code will train a linear SVM on the dataset and plot the
% decision boundary learned.
%
% Load from ex6data1:
% You will have X, y in your environment
load('ex6data1.mat');
fprintf('\nTraining Linear SVM ...\n')
% You should try to change the C value below and see how the decision
% boundary varies (e.g., try C = 1000)
C = 1;
model = svmTrain(X, y, C, @linearKernel, 1e-3, 20);
visualizeBoundaryLinear(X, y, model);
fprintf('Program paused. Press enter to continue.\n');
pause;
%% =============== Part 3: Implementing Gaussian Kernel ===============
% You will now implement the Gaussian kernel to use
% with the SVM. You should complete the code in gaussianKernel.m
%
fprintf('\nEvaluating the Gaussian Kernel ...\n')
x1 = [1 2 1]; x2 = [0 4 -1]; sigma = 2;
sim = gaussianKernel(x1, x2, sigma);
fprintf(['Gaussian Kernel between x1 = [1; 2; 1], x2 = [0; 4; -1], sigma = 0.5 :' ...
'\n\t%f\n(this value should be about 0.324652)\n'], sim);
fprintf('Program paused. Press enter to continue.\n');
pause;
%% =============== Part 4: Visualizing Dataset 2 ================
% The following code will load the next dataset into your environment and
% plot the data.
%
fprintf('Loading and Visualizing Data ...\n')
% Load from ex6data2:
% You will have X, y in your environment
load('ex6data2.mat');
% Plot training data
plotData(X, y);
fprintf('Program paused. Press enter to continue.\n');
pause;
%% ========== Part 5: Training SVM with RBF Kernel (Dataset 2) ==========
% After you have implemented the kernel, we can now use it to train the
% SVM classifier.
%
fprintf('\nTraining SVM with RBF Kernel (this may take 1 to 2 minutes) ...\n');
% Load from ex6data2:
% You will have X, y in your environment
load('ex6data2.mat');
% SVM Parameters
C = 1; sigma = 0.1;
% We set the tolerance and max_passes lower here so that the code will run
% faster. However, in practice, you will want to run the training to
% convergence.
model= svmTrain(X, y, C, @(x1, x2) gaussianKernel(x1, x2, sigma));
visualizeBoundary(X, y, model);
fprintf('Program paused. Press enter to continue.\n');
pause;
%% =============== Part 6: Visualizing Dataset 3 ================
% The following code will load the next dataset into your environment and
% plot the data.
%
fprintf('Loading and Visualizing Data ...\n')
% Load from ex6data3:
% You will have X, y in your environment
load('ex6data3.mat');
% Plot training data
plotData(X, y);
fprintf('Program paused. Press enter to continue.\n');
pause;
%% ========== Part 7: Training SVM with RBF Kernel (Dataset 3) ==========
% This is a different dataset that you can use to experiment with. Try
% different values of C and sigma here.
%
% Load from ex6data3:
% You will have X, y in your environment
load('ex6data3.mat');
% Try different SVM Parameters here
[C, sigma] = dataset3Params(X, y, Xval, yval);
% Train the SVM
model= svmTrain(X, y, C, @(x1, x2) gaussianKernel(x1, x2, sigma));
visualizeBoundary(X, y, model);
fprintf('Program paused. Press enter to continue.\n');
pause;

138
machine-learning-ex6/ex6/ex6_spam.m Normal file
View File

@@ -0,0 +1,138 @@
%% Machine Learning Online Class
% Exercise 6 | Spam Classification with SVMs
%
% Instructions
% ------------
%
% This file contains code that helps you get started on the
% exercise. You will need to complete the following functions:
%
% gaussianKernel.m
% dataset3Params.m
% processEmail.m
% emailFeatures.m
%
% For this exercise, you will not need to change any code in this file,
% or any other files other than those mentioned above.
%
%% Initialization
clear ; close all; clc
%% ==================== Part 1: Email Preprocessing ====================
% To use an SVM to classify emails into Spam v.s. Non-Spam, you first need
% to convert each email into a vector of features. In this part, you will
% implement the preprocessing steps for each email. You should
% complete the code in processEmail.m to produce a word indices vector
% for a given email.
fprintf('\nPreprocessing sample email (emailSample1.txt)\n');
% Extract Features
file_contents = readFile('emailSample1.txt');
word_indices = processEmail(file_contents);
% Print Stats
fprintf('Word Indices: \n');
fprintf(' %d', word_indices);
fprintf('\n\n');
fprintf('Program paused. Press enter to continue.\n');
pause;
%% ==================== Part 2: Feature Extraction ====================
% Now, you will convert each email into a vector of features in R^n.
% You should complete the code in emailFeatures.m to produce a feature
% vector for a given email.
fprintf('\nExtracting features from sample email (emailSample1.txt)\n');
% Extract Features
file_contents = readFile('emailSample1.txt');
word_indices = processEmail(file_contents);
features = emailFeatures(word_indices);
% Print Stats
fprintf('Length of feature vector: %d\n', length(features));
fprintf('Number of non-zero entries: %d\n', sum(features > 0));
fprintf('Program paused. Press enter to continue.\n');
pause;
%% =========== Part 3: Train Linear SVM for Spam Classification ========
% In this section, you will train a linear classifier to determine if an
% email is Spam or Not-Spam.
% Load the Spam Email dataset
% You will have X, y in your environment
load('spamTrain.mat');
fprintf('\nTraining Linear SVM (Spam Classification)\n')
fprintf('(this may take 1 to 2 minutes) ...\n')
C = 0.1;
model = svmTrain(X, y, C, @linearKernel);
p = svmPredict(model, X);
fprintf('Training Accuracy: %f\n', mean(double(p == y)) * 100);
%% =================== Part 4: Test Spam Classification ================
% After training the classifier, we can evaluate it on a test set. We have
% included a test set in spamTest.mat
% Load the test dataset
% You will have Xtest, ytest in your environment
load('spamTest.mat');
fprintf('\nEvaluating the trained Linear SVM on a test set ...\n')
p = svmPredict(model, Xtest);
fprintf('Test Accuracy: %f\n', mean(double(p == ytest)) * 100);
pause;
%% ================= Part 5: Top Predictors of Spam ====================
% Since the model we are training is a linear SVM, we can inspect the
% weights learned by the model to understand better how it is determining
% whether an email is spam or not. The following code finds the words with
% the highest weights in the classifier. Informally, the classifier
% 'thinks' that these words are the most likely indicators of spam.
%
% Sort the weights and obtin the vocabulary list
[weight, idx] = sort(model.w, 'descend');
vocabList = getVocabList();
fprintf('\nTop predictors of spam: \n');
for i = 1:15
fprintf(' %-15s (%f) \n', vocabList{idx(i)}, weight(i));
end
fprintf('\n\n');
fprintf('\nProgram paused. Press enter to continue.\n');
pause;
%% =================== Part 6: Try Your Own Emails =====================
% Now that you've trained the spam classifier, you can use it on your own
% emails! In the starter code, we have included spamSample1.txt,
% spamSample2.txt, emailSample1.txt and emailSample2.txt as examples.
% The following code reads in one of these emails and then uses your
% learned SVM classifier to determine whether the email is Spam or
% Not Spam
% Set the file to be read in (change this to spamSample2.txt,
% emailSample1.txt or emailSample2.txt to see different predictions on
% different emails types). Try your own emails as well!
filename = 'spamSample1.txt';
% Read and predict
file_contents = readFile(filename);
word_indices = processEmail(file_contents);
x = emailFeatures(word_indices);
p = svmPredict(model, x);
fprintf('\nProcessed %s\n\nSpam Classification: %d\n', filename, p);
fprintf('(1 indicates spam, 0 indicates not spam)\n\n');

BIN
machine-learning-ex6/ex6/ex6data1.mat Normal file
View File

Binary file not shown.

BIN
machine-learning-ex6/ex6/ex6data2.mat Normal file
View File

Binary file not shown.

BIN
machine-learning-ex6/ex6/ex6data3.mat Normal file
View File

Binary file not shown.

26
machine-learning-ex6/ex6/gaussianKernel.m Normal file
View File

@@ -0,0 +1,26 @@
function sim = gaussianKernel(x1, x2, sigma)
%RBFKERNEL returns a radial basis function kernel between x1 and x2
% sim = gaussianKernel(x1, x2) returns a gaussian kernel between x1 and x2
% and returns the value in sim
% Ensure that x1 and x2 are column vectors
x1 = x1(:); x2 = x2(:);
% You need to return the following variables correctly.
sim = 0;
% ====================== YOUR CODE HERE ======================
% Instructions: Fill in this function to return the similarity between x1
% and x2 computed using a Gaussian kernel with bandwidth
% sigma
%
%
% =============================================================
end

25
machine-learning-ex6/ex6/getVocabList.m Normal file
View File

@@ -0,0 +1,25 @@
function vocabList = getVocabList()
%GETVOCABLIST reads the fixed vocabulary list in vocab.txt and returns a
%cell array of the words
% vocabList = GETVOCABLIST() reads the fixed vocabulary list in vocab.txt
% and returns a cell array of the words in vocabList.
%% Read the fixed vocabulary list
fid = fopen('vocab.txt');
% Store all dictionary words in cell array vocab{}
n = 1899; % Total number of words in the dictionary
% For ease of implementation, we use a struct to map the strings => integers
% In practice, you'll want to use some form of hashmap
vocabList = cell(n, 1);
for i = 1:n
% Word Index (can ignore since it will be = i)
fscanf(fid, '%d', 1);
% Actual Word
vocabList{i} = fscanf(fid, '%s', 1);
end
fclose(fid);
end

41
machine-learning-ex6/ex6/lib/jsonlab/AUTHORS.txt Normal file
View File

@@ -0,0 +1,41 @@
The author of "jsonlab" toolbox is Qianqian Fang. Qianqian
is currently an Assistant Professor at Massachusetts General Hospital,
Harvard Medical School.
Address: Martinos Center for Biomedical Imaging,
Massachusetts General Hospital,
Harvard Medical School
Bldg 149, 13th St, Charlestown, MA 02129, USA
URL: http://nmr.mgh.harvard.edu/~fangq/
Email: <fangq at nmr.mgh.harvard.edu> or <fangqq at gmail.com>
The script loadjson.m was built upon previous works by
- Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713
date: 2009/11/02
- François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393
date: 2009/03/22
- Joel Feenstra: http://www.mathworks.com/matlabcentral/fileexchange/20565
date: 2008/07/03
This toolbox contains patches submitted by the following contributors:
- Blake Johnson <bjohnso at bbn.com>
part of revision 341
- Niclas Borlin <Niclas.Borlin at cs.umu.se>
various fixes in revision 394, including
- loadjson crashes for all-zero sparse matrix.
- loadjson crashes for empty sparse matrix.
- Non-zero size of 0-by-N and N-by-0 empty matrices is lost after savejson/loadjson.
- loadjson crashes for sparse real column vector.
- loadjson crashes for sparse complex column vector.
- Data is corrupted by savejson for sparse real row vector.
- savejson crashes for sparse complex row vector.
- Yul Kang <yul.kang.on at gmail.com>
patches for svn revision 415.
- savejson saves an empty cell array as [] instead of null
- loadjson differentiates an empty struct from an empty array

74
machine-learning-ex6/ex6/lib/jsonlab/ChangeLog.txt Normal file
View File

@@ -0,0 +1,74 @@
============================================================================
JSONlab - a toolbox to encode/decode JSON/UBJSON files in MATLAB/Octave
----------------------------------------------------------------------------
JSONlab ChangeLog (key features marked by *):
== JSONlab 1.0 (codename: Optimus - Final), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2015/01/02 polish help info for all major functions, update examples, finalize 1.0
2014/12/19 fix a bug to strictly respect NoRowBracket in savejson
== JSONlab 1.0.0-RC2 (codename: Optimus - RC2), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2014/11/22 show progress bar in loadjson ('ShowProgress')
2014/11/17 add Compact option in savejson to output compact JSON format ('Compact')
2014/11/17 add FastArrayParser in loadjson to specify fast parser applicable levels
2014/09/18 start official github mirror: https://github.com/fangq/jsonlab
== JSONlab 1.0.0-RC1 (codename: Optimus - RC1), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2014/09/17 fix several compatibility issues when running on octave versions 3.2-3.8
2014/09/17 support 2D cell and struct arrays in both savejson and saveubjson
2014/08/04 escape special characters in a JSON string
2014/02/16 fix a bug when saving ubjson files
== JSONlab 0.9.9 (codename: Optimus - beta), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2014/01/22 use binary read and write in saveubjson and loadubjson
== JSONlab 0.9.8-1 (codename: Optimus - alpha update 1), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2013/10/07 better round-trip conservation for empty arrays and structs (patch submitted by Yul Kang)
== JSONlab 0.9.8 (codename: Optimus - alpha), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2013/08/23 *universal Binary JSON (UBJSON) support, including both saveubjson and loadubjson
== JSONlab 0.9.1 (codename: Rodimus, update 1), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2012/12/18 *handling of various empty and sparse matrices (fixes submitted by Niclas Borlin)
== JSONlab 0.9.0 (codename: Rodimus), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2012/06/17 *new format for an invalid leading char, unpacking hex code in savejson
2012/06/01 support JSONP in savejson
2012/05/25 fix the empty cell bug (reported by Cyril Davin)
2012/04/05 savejson can save to a file (suggested by Patrick Rapin)
== JSONlab 0.8.1 (codename: Sentiel, Update 1), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2012/02/28 loadjson quotation mark escape bug, see http://bit.ly/yyk1nS
2012/01/25 patch to handle root-less objects, contributed by Blake Johnson
== JSONlab 0.8.0 (codename: Sentiel), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2012/01/13 *speed up loadjson by 20 fold when parsing large data arrays in matlab
2012/01/11 remove row bracket if an array has 1 element, suggested by Mykel Kochenderfer
2011/12/22 *accept sequence of 'param',value input in savejson and loadjson
2011/11/18 fix struct array bug reported by Mykel Kochenderfer
== JSONlab 0.5.1 (codename: Nexus Update 1), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2011/10/21 fix a bug in loadjson, previous code does not use any of the acceleration
2011/10/20 loadjson supports JSON collections - concatenated JSON objects
== JSONlab 0.5.0 (codename: Nexus), FangQ <fangq (at) nmr.mgh.harvard.edu> ==
2011/10/16 package and release jsonlab 0.5.0
2011/10/15 *add json demo and regression test, support cpx numbers, fix double quote bug
2011/10/11 *speed up readjson dramatically, interpret _Array* tags, show data in root level
2011/10/10 create jsonlab project, start jsonlab website, add online documentation
2011/10/07 *speed up savejson by 25x using sprintf instead of mat2str, add options support
2011/10/06 *savejson works for structs, cells and arrays
2011/09/09 derive loadjson from JSON parser from MATLAB Central, draft savejson.m

25
machine-learning-ex6/ex6/lib/jsonlab/LICENSE_BSD.txt Normal file
View File

@@ -0,0 +1,25 @@
Copyright 2011-2015 Qianqian Fang <fangq at nmr.mgh.harvard.edu>. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are
permitted provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list of
conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this list
of conditions and the following disclaimer in the documentation and/or other materials
provided with the distribution.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ''AS IS'' AND ANY EXPRESS OR IMPLIED
WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
The views and conclusions contained in the software and documentation are those of the
authors and should not be interpreted as representing official policies, either expressed
or implied, of the copyright holders.

394
machine-learning-ex6/ex6/lib/jsonlab/README.txt Normal file
View File

@@ -0,0 +1,394 @@
===============================================================================
= JSONLab =
= An open-source MATLAB/Octave JSON encoder and decoder =
===============================================================================
*Copyright (C) 2011-2015 Qianqian Fang <fangq at nmr.mgh.harvard.edu>
*License: BSD License, see License_BSD.txt for details
*Version: 1.0 (Optimus - Final)
-------------------------------------------------------------------------------
Table of Content:
I. Introduction
II. Installation
III.Using JSONLab
IV. Known Issues and TODOs
V. Contribution and feedback
-------------------------------------------------------------------------------
I. Introduction
JSON ([http://www.json.org/ JavaScript Object Notation]) is a highly portable,
human-readable and "[http://en.wikipedia.org/wiki/JSON fat-free]" text format
to represent complex and hierarchical data. It is as powerful as
[http://en.wikipedia.org/wiki/XML XML], but less verbose. JSON format is widely
used for data-exchange in applications, and is essential for the wild success
of [http://en.wikipedia.org/wiki/Ajax_(programming) Ajax] and
[http://en.wikipedia.org/wiki/Web_2.0 Web2.0].
UBJSON (Universal Binary JSON) is a binary JSON format, specifically
optimized for compact file size and better performance while keeping
the semantics as simple as the text-based JSON format. Using the UBJSON
format allows to wrap complex binary data in a flexible and extensible
structure, making it possible to process complex and large dataset
without accuracy loss due to text conversions.
We envision that both JSON and its binary version will serve as part of
the mainstream data-exchange formats for scientific research in the future.
It will provide the flexibility and generality achieved by other popular
general-purpose file specifications, such as
[http://www.hdfgroup.org/HDF5/whatishdf5.html HDF5], with significantly
reduced complexity and enhanced performance.
JSONLab is a free and open-source implementation of a JSON/UBJSON encoder
and a decoder in the native MATLAB language. It can be used to convert a MATLAB
data structure (array, struct, cell, struct array and cell array) into
JSON/UBJSON formatted strings, or to decode a JSON/UBJSON file into MATLAB
data structure. JSONLab supports both MATLAB and
[http://www.gnu.org/software/octave/ GNU Octave] (a free MATLAB clone).
-------------------------------------------------------------------------------
II. Installation
The installation of JSONLab is no different than any other simple
MATLAB toolbox. You only need to download/unzip the JSONLab package
to a folder, and add the folder's path to MATLAB/Octave's path list
by using the following command:
addpath('/path/to/jsonlab');
If you want to add this path permanently, you need to type "pathtool",
browse to the jsonlab root folder and add to the list, then click "Save".
Then, run "rehash" in MATLAB, and type "which loadjson", if you see an
output, that means JSONLab is installed for MATLAB/Octave.
-------------------------------------------------------------------------------
III.Using JSONLab
JSONLab provides two functions, loadjson.m -- a MATLAB->JSON decoder,
and savejson.m -- a MATLAB->JSON encoder, for the text-based JSON, and
two equivallent functions -- loadubjson and saveubjson for the binary
JSON. The detailed help info for the four functions can be found below:
=== loadjson.m ===
<pre>
data=loadjson(fname,opt)
or
data=loadjson(fname,'param1',value1,'param2',value2,...)
parse a JSON (JavaScript Object Notation) file or string
authors:Qianqian Fang (fangq<at> nmr.mgh.harvard.edu)
created on 2011/09/09, including previous works from
Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713
created on 2009/11/02
François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393
created on 2009/03/22
Joel Feenstra:
http://www.mathworks.com/matlabcentral/fileexchange/20565
created on 2008/07/03
$Id: loadjson.m 452 2014-11-22 16:43:33Z fangq $
input:
fname: input file name, if fname contains "{}" or "[]", fname
will be interpreted as a JSON string
opt: a struct to store parsing options, opt can be replaced by
a list of ('param',value) pairs - the param string is equivallent
to a field in opt. opt can have the following
fields (first in [.|.] is the default)
opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat
for each element of the JSON data, and group
arrays based on the cell2mat rules.
opt.FastArrayParser [1|0 or integer]: if set to 1, use a
speed-optimized array parser when loading an
array object. The fast array parser may
collapse block arrays into a single large
array similar to rules defined in cell2mat; 0 to
use a legacy parser; if set to a larger-than-1
value, this option will specify the minimum
dimension to enable the fast array parser. For
example, if the input is a 3D array, setting
FastArrayParser to 1 will return a 3D array;
setting to 2 will return a cell array of 2D
arrays; setting to 3 will return to a 2D cell
array of 1D vectors; setting to 4 will return a
3D cell array.
opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.
output:
dat: a cell array, where {...} blocks are converted into cell arrays,
and [...] are converted to arrays
examples:
dat=loadjson('{"obj":{"string":"value","array":[1,2,3]}}')
dat=loadjson(['examples' filesep 'example1.json'])
dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)
</pre>
=== savejson.m ===
<pre>
json=savejson(rootname,obj,filename)
or
json=savejson(rootname,obj,opt)
json=savejson(rootname,obj,'param1',value1,'param2',value2,...)
convert a MATLAB object (cell, struct or array) into a JSON (JavaScript
Object Notation) string
author: Qianqian Fang (fangq<at> nmr.mgh.harvard.edu)
created on 2011/09/09
$Id: savejson.m 458 2014-12-19 22:17:17Z fangq $
input:
rootname: the name of the root-object, when set to '', the root name
is ignored, however, when opt.ForceRootName is set to 1 (see below),
the MATLAB variable name will be used as the root name.
obj: a MATLAB object (array, cell, cell array, struct, struct array).
filename: a string for the file name to save the output JSON data.
opt: a struct for additional options, ignore to use default values.
opt can have the following fields (first in [.|.] is the default)
opt.FileName [''|string]: a file name to save the output JSON data
opt.FloatFormat ['%.10g'|string]: format to show each numeric element
of a 1D/2D array;
opt.ArrayIndent [1|0]: if 1, output explicit data array with
precedent indentation; if 0, no indentation
opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D
array in JSON array format; if sets to 1, an
array will be shown as a struct with fields
"_ArrayType_", "_ArraySize_" and "_ArrayData_"; for
sparse arrays, the non-zero elements will be
saved to _ArrayData_ field in triplet-format i.e.
(ix,iy,val) and "_ArrayIsSparse_" will be added
with a value of 1; for a complex array, the
_ArrayData_ array will include two columns
(4 for sparse) to record the real and imaginary
parts, and also "_ArrayIsComplex_":1 is added.
opt.ParseLogical [0|1]: if this is set to 1, logical array elem
will use true/false rather than 1/0.
opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single
numerical element will be shown without a square
bracket, unless it is the root object; if 0, square
brackets are forced for any numerical arrays.
opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson
will use the name of the passed obj variable as the
root object name; if obj is an expression and
does not have a name, 'root' will be used; if this
is set to 0 and rootname is empty, the root level
will be merged down to the lower level.
opt.Inf ['"$1_Inf_"'|string]: a customized regular expression pattern
to represent +/-Inf. The matched pattern is '([-+]*)Inf'
and $1 represents the sign. For those who want to use
1e999 to represent Inf, they can set opt.Inf to '$11e999'
opt.NaN ['"_NaN_"'|string]: a customized regular expression pattern
to represent NaN
opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),
for example, if opt.JSONP='foo', the JSON data is
wrapped inside a function call as 'foo(...);'
opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson
back to the string form
opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.
opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)
opt can be replaced by a list of ('param',value) pairs. The param
string is equivallent to a field in opt and is case sensitive.
output:
json: a string in the JSON format (see http://json.org)
examples:
jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],...
'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...
'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...
2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...
'MeshCreator','FangQ','MeshTitle','T6 Cube',...
'SpecialData',[nan, inf, -inf]);
savejson('jmesh',jsonmesh)
savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\t%.5g')
</pre>
=== loadubjson.m ===
<pre>
data=loadubjson(fname,opt)
or
data=loadubjson(fname,'param1',value1,'param2',value2,...)
parse a JSON (JavaScript Object Notation) file or string
authors:Qianqian Fang (fangq<at> nmr.mgh.harvard.edu)
created on 2013/08/01
$Id: loadubjson.m 436 2014-08-05 20:51:40Z fangq $
input:
fname: input file name, if fname contains "{}" or "[]", fname
will be interpreted as a UBJSON string
opt: a struct to store parsing options, opt can be replaced by
a list of ('param',value) pairs - the param string is equivallent
to a field in opt. opt can have the following
fields (first in [.|.] is the default)
opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat
for each element of the JSON data, and group
arrays based on the cell2mat rules.
opt.IntEndian [B|L]: specify the endianness of the integer fields
in the UBJSON input data. B - Big-Endian format for
integers (as required in the UBJSON specification);
L - input integer fields are in Little-Endian order.
output:
dat: a cell array, where {...} blocks are converted into cell arrays,
and [...] are converted to arrays
examples:
obj=struct('string','value','array',[1 2 3]);
ubjdata=saveubjson('obj',obj);
dat=loadubjson(ubjdata)
dat=loadubjson(['examples' filesep 'example1.ubj'])
dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)
</pre>
=== saveubjson.m ===
<pre>
json=saveubjson(rootname,obj,filename)
or
json=saveubjson(rootname,obj,opt)
json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)
convert a MATLAB object (cell, struct or array) into a Universal
Binary JSON (UBJSON) binary string
author: Qianqian Fang (fangq<at> nmr.mgh.harvard.edu)
created on 2013/08/17
$Id: saveubjson.m 440 2014-09-17 19:59:45Z fangq $
input:
rootname: the name of the root-object, when set to '', the root name
is ignored, however, when opt.ForceRootName is set to 1 (see below),
the MATLAB variable name will be used as the root name.
obj: a MATLAB object (array, cell, cell array, struct, struct array)
filename: a string for the file name to save the output UBJSON data
opt: a struct for additional options, ignore to use default values.
opt can have the following fields (first in [.|.] is the default)
opt.FileName [''|string]: a file name to save the output JSON data
opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D
array in JSON array format; if sets to 1, an
array will be shown as a struct with fields
"_ArrayType_", "_ArraySize_" and "_ArrayData_"; for
sparse arrays, the non-zero elements will be
saved to _ArrayData_ field in triplet-format i.e.
(ix,iy,val) and "_ArrayIsSparse_" will be added
with a value of 1; for a complex array, the
_ArrayData_ array will include two columns
(4 for sparse) to record the real and imaginary
parts, and also "_ArrayIsComplex_":1 is added.
opt.ParseLogical [1|0]: if this is set to 1, logical array elem
will use true/false rather than 1/0.
opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single
numerical element will be shown without a square
bracket, unless it is the root object; if 0, square
brackets are forced for any numerical arrays.
opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson
will use the name of the passed obj variable as the
root object name; if obj is an expression and
does not have a name, 'root' will be used; if this
is set to 0 and rootname is empty, the root level
will be merged down to the lower level.
opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),
for example, if opt.JSON='foo', the JSON data is
wrapped inside a function call as 'foo(...);'
opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson
back to the string form
opt can be replaced by a list of ('param',value) pairs. The param
string is equivallent to a field in opt and is case sensitive.
output:
json: a binary string in the UBJSON format (see http://ubjson.org)
examples:
jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],...
'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...
'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...
2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...
'MeshCreator','FangQ','MeshTitle','T6 Cube',...
'SpecialData',[nan, inf, -inf]);
saveubjson('jsonmesh',jsonmesh)
saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')
</pre>
=== examples ===
Under the "examples" folder, you can find several scripts to demonstrate the
basic utilities of JSONLab. Running the "demo_jsonlab_basic.m" script, you
will see the conversions from MATLAB data structure to JSON text and backward.
In "jsonlab_selftest.m", we load complex JSON files downloaded from the Internet
and validate the loadjson/savejson functions for regression testing purposes.
Similarly, a "demo_ubjson_basic.m" script is provided to test the saveubjson
and loadubjson pairs for various matlab data structures.
Please run these examples and understand how JSONLab works before you use
it to process your data.
-------------------------------------------------------------------------------
IV. Known Issues and TODOs
JSONLab has several known limitations. We are striving to make it more general
and robust. Hopefully in a few future releases, the limitations become less.
Here are the known issues:
# 3D or higher dimensional cell/struct-arrays will be converted to 2D arrays;
# When processing names containing multi-byte characters, Octave and MATLAB \
can give different field-names; you can use feature('DefaultCharacterSet','latin1') \
in MATLAB to get consistant results
# savejson can not handle class and dataset.
# saveubjson converts a logical array into a uint8 ([U]) array
# an unofficial N-D array count syntax is implemented in saveubjson. We are \
actively communicating with the UBJSON spec maintainer to investigate the \
possibility of making it upstream
# loadubjson can not parse all UBJSON Specification (Draft 9) compliant \
files, however, it can parse all UBJSON files produced by saveubjson.
-------------------------------------------------------------------------------
V. Contribution and feedback
JSONLab is an open-source project. This means you can not only use it and modify
it as you wish, but also you can contribute your changes back to JSONLab so
that everyone else can enjoy the improvement. For anyone who want to contribute,
please download JSONLab source code from it's subversion repository by using the
following command:
svn checkout svn://svn.code.sf.net/p/iso2mesh/code/trunk/jsonlab jsonlab
You can make changes to the files as needed. Once you are satisfied with your
changes, and ready to share it with others, please cd the root directory of
JSONLab, and type
svn diff > yourname_featurename.patch
You then email the .patch file to JSONLab's maintainer, Qianqian Fang, at
the email address shown in the beginning of this file. Qianqian will review
the changes and commit it to the subversion if they are satisfactory.
We appreciate any suggestions and feedbacks from you. Please use iso2mesh's
mailing list to report any questions you may have with JSONLab:
http://groups.google.com/group/iso2mesh-users?hl=en&pli=1
(Subscription to the mailing list is needed in order to post messages).

32
machine-learning-ex6/ex6/lib/jsonlab/jsonopt.m Normal file
View File

@@ -0,0 +1,32 @@
function val=jsonopt(key,default,varargin)
%
% val=jsonopt(key,default,optstruct)
%
% setting options based on a struct. The struct can be produced
% by varargin2struct from a list of 'param','value' pairs
%
% authors:Qianqian Fang (fangq<at> nmr.mgh.harvard.edu)
%
% $Id: loadjson.m 371 2012-06-20 12:43:06Z fangq $
%
% input:
% key: a string with which one look up a value from a struct
% default: if the key does not exist, return default
% optstruct: a struct where each sub-field is a key
%
% output:
% val: if key exists, val=optstruct.key; otherwise val=default
%
% license:
% BSD, see LICENSE_BSD.txt files for details
%
% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)
%
val=default;
if(nargin<=2) return; end
opt=varargin{1};
if(isstruct(opt) && isfield(opt,key))
val=getfield(opt,key);
end

566
machine-learning-ex6/ex6/lib/jsonlab/loadjson.m Normal file
View File

@@ -0,0 +1,566 @@
function data = loadjson(fname,varargin)
%
% data=loadjson(fname,opt)
% or
% data=loadjson(fname,'param1',value1,'param2',value2,...)
%
% parse a JSON (JavaScript Object Notation) file or string
%
% authors:Qianqian Fang (fangq<at> nmr.mgh.harvard.edu)
% created on 2011/09/09, including previous works from
%
% Nedialko Krouchev: http://www.mathworks.com/matlabcentral/fileexchange/25713
% created on 2009/11/02
% François Glineur: http://www.mathworks.com/matlabcentral/fileexchange/23393
% created on 2009/03/22
% Joel Feenstra:
% http://www.mathworks.com/matlabcentral/fileexchange/20565
% created on 2008/07/03
%
% $Id: loadjson.m 460 2015-01-03 00:30:45Z fangq $
%
% input:
% fname: input file name, if fname contains "{}" or "[]", fname
% will be interpreted as a JSON string
% opt: a struct to store parsing options, opt can be replaced by
% a list of ('param',value) pairs - the param string is equivallent
% to a field in opt. opt can have the following
% fields (first in [.|.] is the default)
%
% opt.SimplifyCell [0|1]: if set to 1, loadjson will call cell2mat
% for each element of the JSON data, and group
% arrays based on the cell2mat rules.
% opt.FastArrayParser [1|0 or integer]: if set to 1, use a
% speed-optimized array parser when loading an
% array object. The fast array parser may
% collapse block arrays into a single large
% array similar to rules defined in cell2mat; 0 to
% use a legacy parser; if set to a larger-than-1
% value, this option will specify the minimum
% dimension to enable the fast array parser. For
% example, if the input is a 3D array, setting
% FastArrayParser to 1 will return a 3D array;
% setting to 2 will return a cell array of 2D
% arrays; setting to 3 will return to a 2D cell
% array of 1D vectors; setting to 4 will return a
% 3D cell array.
% opt.ShowProgress [0|1]: if set to 1, loadjson displays a progress bar.
%
% output:
% dat: a cell array, where {...} blocks are converted into cell arrays,
% and [...] are converted to arrays
%
% examples:
% dat=loadjson('{"obj":{"string":"value","array":[1,2,3]}}')
% dat=loadjson(['examples' filesep 'example1.json'])
% dat=loadjson(['examples' filesep 'example1.json'],'SimplifyCell',1)
%
% license:
% BSD, see LICENSE_BSD.txt files for details
%
% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)
%
global pos inStr len esc index_esc len_esc isoct arraytoken
if(regexp(fname,'[\{\}\]\[]','once'))
string=fname;
elseif(exist(fname,'file'))
fid = fopen(fname,'rb');
string = fread(fid,inf,'uint8=>char')';
fclose(fid);
else
error('input file does not exist');
end
pos = 1; len = length(string); inStr = string;
isoct=exist('OCTAVE_VERSION','builtin');
arraytoken=find(inStr=='[' | inStr==']' | inStr=='"');
jstr=regexprep(inStr,'\\\\',' ');
escquote=regexp(jstr,'\\"');
arraytoken=sort([arraytoken escquote]);
% String delimiters and escape chars identified to improve speed:
esc = find(inStr=='"' | inStr=='\' ); % comparable to: regexp(inStr, '["\\]');
index_esc = 1; len_esc = length(esc);
opt=varargin2struct(varargin{:});
if(jsonopt('ShowProgress',0,opt)==1)
opt.progressbar_=waitbar(0,'loading ...');
end
jsoncount=1;
while pos <= len
switch(next_char)
case '{'
data{jsoncount} = parse_object(opt);
case '['
data{jsoncount} = parse_array(opt);
otherwise
error_pos('Outer level structure must be an object or an array');
end
jsoncount=jsoncount+1;
end % while
jsoncount=length(data);
if(jsoncount==1 && iscell(data))
data=data{1};
end
if(~isempty(data))
if(isstruct(data)) % data can be a struct array
data=jstruct2array(data);
elseif(iscell(data))
data=jcell2array(data);
end
end
if(isfield(opt,'progressbar_'))
close(opt.progressbar_);
end
%%
function newdata=jcell2array(data)
len=length(data);
newdata=data;
for i=1:len
if(isstruct(data{i}))
newdata{i}=jstruct2array(data{i});
elseif(iscell(data{i}))
newdata{i}=jcell2array(data{i});
end
end
%%-------------------------------------------------------------------------
function newdata=jstruct2array(data)
fn=fieldnames(data);
newdata=data;
len=length(data);
for i=1:length(fn) % depth-first
for j=1:len
if(isstruct(getfield(data(j),fn{i})))
newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));
end
end
end
if(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))
newdata=cell(len,1);
for j=1:len
ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);
iscpx=0;
if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))
if(data(j).x0x5F_ArrayIsComplex_)
iscpx=1;
end
end
if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))
if(data(j).x0x5F_ArrayIsSparse_)
if(~isempty(strmatch('x0x5F_ArraySize_',fn)))
dim=data(j).x0x5F_ArraySize_;
if(iscpx && size(ndata,2)==4-any(dim==1))
ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));
end
if isempty(ndata)
% All-zeros sparse
ndata=sparse(dim(1),prod(dim(2:end)));
elseif dim(1)==1
% Sparse row vector
ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));
elseif dim(2)==1
% Sparse column vector
ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));
else
% Generic sparse array.
ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));
end
else
if(iscpx && size(ndata,2)==4)
ndata(:,3)=complex(ndata(:,3),ndata(:,4));
end
ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));
end
end
elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))
if(iscpx && size(ndata,2)==2)
ndata=complex(ndata(:,1),ndata(:,2));
end
ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);
end
newdata{j}=ndata;
end
if(len==1)
newdata=newdata{1};
end
end
%%-------------------------------------------------------------------------
function object = parse_object(varargin)
parse_char('{');
object = [];
if next_char ~= '}'
while 1
str = parseStr(varargin{:});
if isempty(str)
error_pos('Name of value at position %d cannot be empty');
end
parse_char(':');
val = parse_value(varargin{:});
eval( sprintf( 'object.%s = val;', valid_field(str) ) );
if next_char == '}'
break;
end
parse_char(',');
end
end
parse_char('}');
%%-------------------------------------------------------------------------
function object = parse_array(varargin) % JSON array is written in row-major order
global pos inStr isoct
parse_char('[');
object = cell(0, 1);
dim2=[];
arraydepth=jsonopt('JSONLAB_ArrayDepth_',1,varargin{:});
pbar=jsonopt('progressbar_',-1,varargin{:});
if next_char ~= ']'
if(jsonopt('FastArrayParser',1,varargin{:})>=1 && arraydepth>=jsonopt('FastArrayParser',1,varargin{:}))
[endpos, e1l, e1r, maxlevel]=matching_bracket(inStr,pos);
arraystr=['[' inStr(pos:endpos)];
arraystr=regexprep(arraystr,'"_NaN_"','NaN');
arraystr=regexprep(arraystr,'"([-+]*)_Inf_"','$1Inf');
arraystr(arraystr==sprintf('\n'))=[];
arraystr(arraystr==sprintf('\r'))=[];
%arraystr=regexprep(arraystr,'\s*,',','); % this is slow,sometimes needed
if(~isempty(e1l) && ~isempty(e1r)) % the array is in 2D or higher D
astr=inStr((e1l+1):(e1r-1));
astr=regexprep(astr,'"_NaN_"','NaN');
astr=regexprep(astr,'"([-+]*)_Inf_"','$1Inf');
astr(astr==sprintf('\n'))=[];
astr(astr==sprintf('\r'))=[];
astr(astr==' ')='';
if(isempty(find(astr=='[', 1))) % array is 2D
dim2=length(sscanf(astr,'%f,',[1 inf]));
end
else % array is 1D
astr=arraystr(2:end-1);
astr(astr==' ')='';
[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',[1,inf]);
if(nextidx>=length(astr)-1)
object=obj;
pos=endpos;
parse_char(']');
return;
end
end
if(~isempty(dim2))
astr=arraystr;
astr(astr=='[')='';
astr(astr==']')='';
astr(astr==' ')='';
[obj, count, errmsg, nextidx]=sscanf(astr,'%f,',inf);
if(nextidx>=length(astr)-1)
object=reshape(obj,dim2,numel(obj)/dim2)';
pos=endpos;
parse_char(']');
if(pbar>0)
waitbar(pos/length(inStr),pbar,'loading ...');
end
return;
end
end
arraystr=regexprep(arraystr,'\]\s*,','];');
else
arraystr='[';
end
try
if(isoct && regexp(arraystr,'"','once'))
error('Octave eval can produce empty cells for JSON-like input');
end
object=eval(arraystr);
pos=endpos;
catch
while 1
newopt=varargin2struct(varargin{:},'JSONLAB_ArrayDepth_',arraydepth+1);
val = parse_value(newopt);
object{end+1} = val;
if next_char == ']'
break;
end
parse_char(',');
end
end
end
if(jsonopt('SimplifyCell',0,varargin{:})==1)
try
oldobj=object;
object=cell2mat(object')';
if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)
object=oldobj;
elseif(size(object,1)>1 && ndims(object)==2)
object=object';
end
catch
end
end
parse_char(']');
if(pbar>0)
waitbar(pos/length(inStr),pbar,'loading ...');
end
%%-------------------------------------------------------------------------
function parse_char(c)
global pos inStr len
skip_whitespace;
if pos > len || inStr(pos) ~= c
error_pos(sprintf('Expected %c at position %%d', c));
else
pos = pos + 1;
skip_whitespace;
end
%%-------------------------------------------------------------------------
function c = next_char
global pos inStr len
skip_whitespace;
if pos > len
c = [];
else
c = inStr(pos);
end
%%-------------------------------------------------------------------------
function skip_whitespace
global pos inStr len
while pos <= len && isspace(inStr(pos))
pos = pos + 1;
end
%%-------------------------------------------------------------------------
function str = parseStr(varargin)
global pos inStr len esc index_esc len_esc
% len, ns = length(inStr), keyboard
if inStr(pos) ~= '"'
error_pos('String starting with " expected at position %d');
else
pos = pos + 1;
end
str = '';
while pos <= len
while index_esc <= len_esc && esc(index_esc) < pos
index_esc = index_esc + 1;
end
if index_esc > len_esc
str = [str inStr(pos:len)];
pos = len + 1;
break;
else
str = [str inStr(pos:esc(index_esc)-1)];
pos = esc(index_esc);
end
nstr = length(str); switch inStr(pos)
case '"'
pos = pos + 1;
if(~isempty(str))
if(strcmp(str,'_Inf_'))
str=Inf;
elseif(strcmp(str,'-_Inf_'))
str=-Inf;
elseif(strcmp(str,'_NaN_'))
str=NaN;
end
end
return;
case '\'
if pos+1 > len
error_pos('End of file reached right after escape character');
end
pos = pos + 1;
switch inStr(pos)
case {'"' '\' '/'}
str(nstr+1) = inStr(pos);
pos = pos + 1;
case {'b' 'f' 'n' 'r' 't'}
str(nstr+1) = sprintf(['\' inStr(pos)]);
pos = pos + 1;
case 'u'
if pos+4 > len
error_pos('End of file reached in escaped unicode character');
end
str(nstr+(1:6)) = inStr(pos-1:pos+4);
pos = pos + 5;
end
otherwise % should never happen
str(nstr+1) = inStr(pos), keyboard
pos = pos + 1;
end
end
error_pos('End of file while expecting end of inStr');
%%-------------------------------------------------------------------------
function num = parse_number(varargin)
global pos inStr len isoct
currstr=inStr(pos:end);
numstr=0;
if(isoct~=0)
numstr=regexp(currstr,'^\s*-?(?:0|[1-9]\d*)(?:\.\d+)?(?:[eE][+\-]?\d+)?','end');
[num, one] = sscanf(currstr, '%f', 1);
delta=numstr+1;
else
[num, one, err, delta] = sscanf(currstr, '%f', 1);
if ~isempty(err)
error_pos('Error reading number at position %d');
end
end
pos = pos + delta-1;
%%-------------------------------------------------------------------------
function val = parse_value(varargin)
global pos inStr len
true = 1; false = 0;
pbar=jsonopt('progressbar_',-1,varargin{:});
if(pbar>0)
waitbar(pos/len,pbar,'loading ...');
end
switch(inStr(pos))
case '"'
val = parseStr(varargin{:});
return;
case '['
val = parse_array(varargin{:});
return;
case '{'
val = parse_object(varargin{:});
if isstruct(val)
if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))
val=jstruct2array(val);
end
elseif isempty(val)
val = struct;
end
return;
case {'-','0','1','2','3','4','5','6','7','8','9'}
val = parse_number(varargin{:});
return;
case 't'
if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'true')
val = true;
pos = pos + 4;
return;
end
case 'f'
if pos+4 <= len && strcmpi(inStr(pos:pos+4), 'false')
val = false;
pos = pos + 5;
return;
end
case 'n'
if pos+3 <= len && strcmpi(inStr(pos:pos+3), 'null')
val = [];
pos = pos + 4;
return;
end
end
error_pos('Value expected at position %d');
%%-------------------------------------------------------------------------
function error_pos(msg)
global pos inStr len
poShow = max(min([pos-15 pos-1 pos pos+20],len),1);
if poShow(3) == poShow(2)
poShow(3:4) = poShow(2)+[0 -1]; % display nothing after
end
msg = [sprintf(msg, pos) ': ' ...
inStr(poShow(1):poShow(2)) '<error>' inStr(poShow(3):poShow(4)) ];
error( ['JSONparser:invalidFormat: ' msg] );
%%-------------------------------------------------------------------------
function str = valid_field(str)
global isoct
% From MATLAB doc: field names must begin with a letter, which may be
% followed by any combination of letters, digits, and underscores.
% Invalid characters will be converted to underscores, and the prefix
% "x0x[Hex code]_" will be added if the first character is not a letter.
pos=regexp(str,'^[^A-Za-z]','once');
if(~isempty(pos))
if(~isoct)
str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');
else
str=sprintf('x0x%X_%s',char(str(1)),str(2:end));
end
end
if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end
if(~isoct)
str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');
else
pos=regexp(str,'[^0-9A-Za-z_]');
if(isempty(pos)) return; end
str0=str;
pos0=[0 pos(:)' length(str)];
str='';
for i=1:length(pos)
str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];
end
if(pos(end)~=length(str))
str=[str str0(pos0(end-1)+1:pos0(end))];
end
end
%str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';
%%-------------------------------------------------------------------------
function endpos = matching_quote(str,pos)
len=length(str);
while(pos<len)
if(str(pos)=='"')
if(~(pos>1 && str(pos-1)=='\'))
endpos=pos;
return;
end
end
pos=pos+1;
end
error('unmatched quotation mark');
%%-------------------------------------------------------------------------
function [endpos, e1l, e1r, maxlevel] = matching_bracket(str,pos)
global arraytoken
level=1;
maxlevel=level;
endpos=0;
bpos=arraytoken(arraytoken>=pos);
tokens=str(bpos);
len=length(tokens);
pos=1;
e1l=[];
e1r=[];
while(pos<=len)
c=tokens(pos);
if(c==']')
level=level-1;
if(isempty(e1r)) e1r=bpos(pos); end
if(level==0)
endpos=bpos(pos);
return
end
end
if(c=='[')
if(isempty(e1l)) e1l=bpos(pos); end
level=level+1;
maxlevel=max(maxlevel,level);
end
if(c=='"')
pos=matching_quote(tokens,pos+1);
end
pos=pos+1;
end
if(endpos==0)
error('unmatched "]"');
end

528
machine-learning-ex6/ex6/lib/jsonlab/loadubjson.m Normal file
View File

@@ -0,0 +1,528 @@
function data = loadubjson(fname,varargin)
%
% data=loadubjson(fname,opt)
% or
% data=loadubjson(fname,'param1',value1,'param2',value2,...)
%
% parse a JSON (JavaScript Object Notation) file or string
%
% authors:Qianqian Fang (fangq<at> nmr.mgh.harvard.edu)
% created on 2013/08/01
%
% $Id: loadubjson.m 460 2015-01-03 00:30:45Z fangq $
%
% input:
% fname: input file name, if fname contains "{}" or "[]", fname
% will be interpreted as a UBJSON string
% opt: a struct to store parsing options, opt can be replaced by
% a list of ('param',value) pairs - the param string is equivallent
% to a field in opt. opt can have the following
% fields (first in [.|.] is the default)
%
% opt.SimplifyCell [0|1]: if set to 1, loadubjson will call cell2mat
% for each element of the JSON data, and group
% arrays based on the cell2mat rules.
% opt.IntEndian [B|L]: specify the endianness of the integer fields
% in the UBJSON input data. B - Big-Endian format for
% integers (as required in the UBJSON specification);
% L - input integer fields are in Little-Endian order.
%
% output:
% dat: a cell array, where {...} blocks are converted into cell arrays,
% and [...] are converted to arrays
%
% examples:
% obj=struct('string','value','array',[1 2 3]);
% ubjdata=saveubjson('obj',obj);
% dat=loadubjson(ubjdata)
% dat=loadubjson(['examples' filesep 'example1.ubj'])
% dat=loadubjson(['examples' filesep 'example1.ubj'],'SimplifyCell',1)
%
% license:
% BSD, see LICENSE_BSD.txt files for details
%
% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)
%
global pos inStr len esc index_esc len_esc isoct arraytoken fileendian systemendian
if(regexp(fname,'[\{\}\]\[]','once'))
string=fname;
elseif(exist(fname,'file'))
fid = fopen(fname,'rb');
string = fread(fid,inf,'uint8=>char')';
fclose(fid);
else
error('input file does not exist');
end
pos = 1; len = length(string); inStr = string;
isoct=exist('OCTAVE_VERSION','builtin');
arraytoken=find(inStr=='[' | inStr==']' | inStr=='"');
jstr=regexprep(inStr,'\\\\',' ');
escquote=regexp(jstr,'\\"');
arraytoken=sort([arraytoken escquote]);
% String delimiters and escape chars identified to improve speed:
esc = find(inStr=='"' | inStr=='\' ); % comparable to: regexp(inStr, '["\\]');
index_esc = 1; len_esc = length(esc);
opt=varargin2struct(varargin{:});
fileendian=upper(jsonopt('IntEndian','B',opt));
[os,maxelem,systemendian]=computer;
jsoncount=1;
while pos <= len
switch(next_char)
case '{'
data{jsoncount} = parse_object(opt);
case '['
data{jsoncount} = parse_array(opt);
otherwise
error_pos('Outer level structure must be an object or an array');
end
jsoncount=jsoncount+1;
end % while
jsoncount=length(data);
if(jsoncount==1 && iscell(data))
data=data{1};
end
if(~isempty(data))
if(isstruct(data)) % data can be a struct array
data=jstruct2array(data);
elseif(iscell(data))
data=jcell2array(data);
end
end
%%
function newdata=parse_collection(id,data,obj)
if(jsoncount>0 && exist('data','var'))
if(~iscell(data))
newdata=cell(1);
newdata{1}=data;
data=newdata;
end
end
%%
function newdata=jcell2array(data)
len=length(data);
newdata=data;
for i=1:len
if(isstruct(data{i}))
newdata{i}=jstruct2array(data{i});
elseif(iscell(data{i}))
newdata{i}=jcell2array(data{i});
end
end
%%-------------------------------------------------------------------------
function newdata=jstruct2array(data)
fn=fieldnames(data);
newdata=data;
len=length(data);
for i=1:length(fn) % depth-first
for j=1:len
if(isstruct(getfield(data(j),fn{i})))
newdata(j)=setfield(newdata(j),fn{i},jstruct2array(getfield(data(j),fn{i})));
end
end
end
if(~isempty(strmatch('x0x5F_ArrayType_',fn)) && ~isempty(strmatch('x0x5F_ArrayData_',fn)))
newdata=cell(len,1);
for j=1:len
ndata=cast(data(j).x0x5F_ArrayData_,data(j).x0x5F_ArrayType_);
iscpx=0;
if(~isempty(strmatch('x0x5F_ArrayIsComplex_',fn)))
if(data(j).x0x5F_ArrayIsComplex_)
iscpx=1;
end
end
if(~isempty(strmatch('x0x5F_ArrayIsSparse_',fn)))
if(data(j).x0x5F_ArrayIsSparse_)
if(~isempty(strmatch('x0x5F_ArraySize_',fn)))
dim=double(data(j).x0x5F_ArraySize_);
if(iscpx && size(ndata,2)==4-any(dim==1))
ndata(:,end-1)=complex(ndata(:,end-1),ndata(:,end));
end
if isempty(ndata)
% All-zeros sparse
ndata=sparse(dim(1),prod(dim(2:end)));
elseif dim(1)==1
% Sparse row vector
ndata=sparse(1,ndata(:,1),ndata(:,2),dim(1),prod(dim(2:end)));
elseif dim(2)==1
% Sparse column vector
ndata=sparse(ndata(:,1),1,ndata(:,2),dim(1),prod(dim(2:end)));
else
% Generic sparse array.
ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3),dim(1),prod(dim(2:end)));
end
else
if(iscpx && size(ndata,2)==4)
ndata(:,3)=complex(ndata(:,3),ndata(:,4));
end
ndata=sparse(ndata(:,1),ndata(:,2),ndata(:,3));
end
end
elseif(~isempty(strmatch('x0x5F_ArraySize_',fn)))
if(iscpx && size(ndata,2)==2)
ndata=complex(ndata(:,1),ndata(:,2));
end
ndata=reshape(ndata(:),data(j).x0x5F_ArraySize_);
end
newdata{j}=ndata;
end
if(len==1)
newdata=newdata{1};
end
end
%%-------------------------------------------------------------------------
function object = parse_object(varargin)
parse_char('{');
object = [];
type='';
count=-1;
if(next_char == '$')
type=inStr(pos+1); % TODO
pos=pos+2;
end
if(next_char == '#')
pos=pos+1;
count=double(parse_number());
end
if next_char ~= '}'
num=0;
while 1
str = parseStr(varargin{:});
if isempty(str)
error_pos('Name of value at position %d cannot be empty');
end
%parse_char(':');
val = parse_value(varargin{:});
num=num+1;
eval( sprintf( 'object.%s = val;', valid_field(str) ) );
if next_char == '}' || (count>=0 && num>=count)
break;
end
%parse_char(',');
end
end
if(count==-1)
parse_char('}');
end
%%-------------------------------------------------------------------------
function [cid,len]=elem_info(type)
id=strfind('iUIlLdD',type);
dataclass={'int8','uint8','int16','int32','int64','single','double'};
bytelen=[1,1,2,4,8,4,8];
if(id>0)
cid=dataclass{id};
len=bytelen(id);
else
error_pos('unsupported type at position %d');
end
%%-------------------------------------------------------------------------
function [data adv]=parse_block(type,count,varargin)
global pos inStr isoct fileendian systemendian
[cid,len]=elem_info(type);
datastr=inStr(pos:pos+len*count-1);
if(isoct)
newdata=int8(datastr);
else
newdata=uint8(datastr);
end
id=strfind('iUIlLdD',type);
if(id<=5 && fileendian~=systemendian)
newdata=swapbytes(typecast(newdata,cid));
end
data=typecast(newdata,cid);
adv=double(len*count);
%%-------------------------------------------------------------------------
function object = parse_array(varargin) % JSON array is written in row-major order
global pos inStr isoct
parse_char('[');
object = cell(0, 1);
dim=[];
type='';
count=-1;
if(next_char == '$')
type=inStr(pos+1);
pos=pos+2;
end
if(next_char == '#')
pos=pos+1;
if(next_char=='[')
dim=parse_array(varargin{:});
count=prod(double(dim));
else
count=double(parse_number());
end
end
if(~isempty(type))
if(count>=0)
[object adv]=parse_block(type,count,varargin{:});
if(~isempty(dim))
object=reshape(object,dim);
end
pos=pos+adv;
return;
else
endpos=matching_bracket(inStr,pos);
[cid,len]=elem_info(type);
count=(endpos-pos)/len;
[object adv]=parse_block(type,count,varargin{:});
pos=pos+adv;
parse_char(']');
return;
end
end
if next_char ~= ']'
while 1
val = parse_value(varargin{:});
object{end+1} = val;
if next_char == ']'
break;
end
%parse_char(',');
end
end
if(jsonopt('SimplifyCell',0,varargin{:})==1)
try
oldobj=object;
object=cell2mat(object')';
if(iscell(oldobj) && isstruct(object) && numel(object)>1 && jsonopt('SimplifyCellArray',1,varargin{:})==0)
object=oldobj;
elseif(size(object,1)>1 && ndims(object)==2)
object=object';
end
catch
end
end
if(count==-1)
parse_char(']');
end
%%-------------------------------------------------------------------------
function parse_char(c)
global pos inStr len
skip_whitespace;
if pos > len || inStr(pos) ~= c
error_pos(sprintf('Expected %c at position %%d', c));
else
pos = pos + 1;
skip_whitespace;
end
%%-------------------------------------------------------------------------
function c = next_char
global pos inStr len
skip_whitespace;
if pos > len
c = [];
else
c = inStr(pos);
end
%%-------------------------------------------------------------------------
function skip_whitespace
global pos inStr len
while pos <= len && isspace(inStr(pos))
pos = pos + 1;
end
%%-------------------------------------------------------------------------
function str = parseStr(varargin)
global pos inStr esc index_esc len_esc
% len, ns = length(inStr), keyboard
type=inStr(pos);
if type ~= 'S' && type ~= 'C' && type ~= 'H'
error_pos('String starting with S expected at position %d');
else
pos = pos + 1;
end
if(type == 'C')
str=inStr(pos);
pos=pos+1;
return;
end
bytelen=double(parse_number());
if(length(inStr)>=pos+bytelen-1)
str=inStr(pos:pos+bytelen-1);
pos=pos+bytelen;
else
error_pos('End of file while expecting end of inStr');
end
%%-------------------------------------------------------------------------
function num = parse_number(varargin)
global pos inStr len isoct fileendian systemendian
id=strfind('iUIlLdD',inStr(pos));
if(isempty(id))
error_pos('expecting a number at position %d');
end
type={'int8','uint8','int16','int32','int64','single','double'};
bytelen=[1,1,2,4,8,4,8];
datastr=inStr(pos+1:pos+bytelen(id));
if(isoct)
newdata=int8(datastr);
else
newdata=uint8(datastr);
end
if(id<=5 && fileendian~=systemendian)
newdata=swapbytes(typecast(newdata,type{id}));
end
num=typecast(newdata,type{id});
pos = pos + bytelen(id)+1;
%%-------------------------------------------------------------------------
function val = parse_value(varargin)
global pos inStr len
true = 1; false = 0;
switch(inStr(pos))
case {'S','C','H'}
val = parseStr(varargin{:});
return;
case '['
val = parse_array(varargin{:});
return;
case '{'
val = parse_object(varargin{:});
if isstruct(val)
if(~isempty(strmatch('x0x5F_ArrayType_',fieldnames(val), 'exact')))
val=jstruct2array(val);
end
elseif isempty(val)
val = struct;
end
return;
case {'i','U','I','l','L','d','D'}
val = parse_number(varargin{:});
return;
case 'T'
val = true;
pos = pos + 1;
return;
case 'F'
val = false;
pos = pos + 1;
return;
case {'Z','N'}
val = [];
pos = pos + 1;
return;
end
error_pos('Value expected at position %d');
%%-------------------------------------------------------------------------
function error_pos(msg)
global pos inStr len
poShow = max(min([pos-15 pos-1 pos pos+20],len),1);
if poShow(3) == poShow(2)
poShow(3:4) = poShow(2)+[0 -1]; % display nothing after
end
msg = [sprintf(msg, pos) ': ' ...
inStr(poShow(1):poShow(2)) '<error>' inStr(poShow(3):poShow(4)) ];
error( ['JSONparser:invalidFormat: ' msg] );
%%-------------------------------------------------------------------------
function str = valid_field(str)
global isoct
% From MATLAB doc: field names must begin with a letter, which may be
% followed by any combination of letters, digits, and underscores.
% Invalid characters will be converted to underscores, and the prefix
% "x0x[Hex code]_" will be added if the first character is not a letter.
pos=regexp(str,'^[^A-Za-z]','once');
if(~isempty(pos))
if(~isoct)
str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');
else
str=sprintf('x0x%X_%s',char(str(1)),str(2:end));
end
end
if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end
if(~isoct)
str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');
else
pos=regexp(str,'[^0-9A-Za-z_]');
if(isempty(pos)) return; end
str0=str;
pos0=[0 pos(:)' length(str)];
str='';
for i=1:length(pos)
str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];
end
if(pos(end)~=length(str))
str=[str str0(pos0(end-1)+1:pos0(end))];
end
end
%str(~isletter(str) & ~('0' <= str & str <= '9')) = '_';
%%-------------------------------------------------------------------------
function endpos = matching_quote(str,pos)
len=length(str);
while(pos<len)
if(str(pos)=='"')
if(~(pos>1 && str(pos-1)=='\'))
endpos=pos;
return;
end
end
pos=pos+1;
end
error('unmatched quotation mark');
%%-------------------------------------------------------------------------
function [endpos e1l e1r maxlevel] = matching_bracket(str,pos)
global arraytoken
level=1;
maxlevel=level;
endpos=0;
bpos=arraytoken(arraytoken>=pos);
tokens=str(bpos);
len=length(tokens);
pos=1;
e1l=[];
e1r=[];
while(pos<=len)
c=tokens(pos);
if(c==']')
level=level-1;
if(isempty(e1r)) e1r=bpos(pos); end
if(level==0)
endpos=bpos(pos);
return
end
end
if(c=='[')
if(isempty(e1l)) e1l=bpos(pos); end
level=level+1;
maxlevel=max(maxlevel,level);
end
if(c=='"')
pos=matching_quote(tokens,pos+1);
end
pos=pos+1;
end
if(endpos==0)
error('unmatched "]"');
end

33
machine-learning-ex6/ex6/lib/jsonlab/mergestruct.m Normal file
View File

@@ -0,0 +1,33 @@
function s=mergestruct(s1,s2)
%
% s=mergestruct(s1,s2)
%
% merge two struct objects into one
%
% authors:Qianqian Fang (fangq<at> nmr.mgh.harvard.edu)
% date: 2012/12/22
%
% input:
% s1,s2: a struct object, s1 and s2 can not be arrays
%
% output:
% s: the merged struct object. fields in s1 and s2 will be combined in s.
%
% license:
% BSD, see LICENSE_BSD.txt files for details
%
% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)
%
if(~isstruct(s1) || ~isstruct(s2))
error('input parameters contain non-struct');
end
if(length(s1)>1 || length(s2)>1)
error('can not merge struct arrays');
end
fn=fieldnames(s2);
s=s1;
for i=1:length(fn)
s=setfield(s,fn{i},getfield(s2,fn{i}));
end

475
machine-learning-ex6/ex6/lib/jsonlab/savejson.m Normal file
View File

@@ -0,0 +1,475 @@
function json=savejson(rootname,obj,varargin)
%
% json=savejson(rootname,obj,filename)
% or
% json=savejson(rootname,obj,opt)
% json=savejson(rootname,obj,'param1',value1,'param2',value2,...)
%
% convert a MATLAB object (cell, struct or array) into a JSON (JavaScript
% Object Notation) string
%
% author: Qianqian Fang (fangq<at> nmr.mgh.harvard.edu)
% created on 2011/09/09
%
% $Id: savejson.m 460 2015-01-03 00:30:45Z fangq $
%
% input:
% rootname: the name of the root-object, when set to '', the root name
% is ignored, however, when opt.ForceRootName is set to 1 (see below),
% the MATLAB variable name will be used as the root name.
% obj: a MATLAB object (array, cell, cell array, struct, struct array).
% filename: a string for the file name to save the output JSON data.
% opt: a struct for additional options, ignore to use default values.
% opt can have the following fields (first in [.|.] is the default)
%
% opt.FileName [''|string]: a file name to save the output JSON data
% opt.FloatFormat ['%.10g'|string]: format to show each numeric element
% of a 1D/2D array;
% opt.ArrayIndent [1|0]: if 1, output explicit data array with
% precedent indentation; if 0, no indentation
% opt.ArrayToStruct[0|1]: when set to 0, savejson outputs 1D/2D
% array in JSON array format; if sets to 1, an
% array will be shown as a struct with fields
% "_ArrayType_", "_ArraySize_" and "_ArrayData_"; for
% sparse arrays, the non-zero elements will be
% saved to _ArrayData_ field in triplet-format i.e.
% (ix,iy,val) and "_ArrayIsSparse_" will be added
% with a value of 1; for a complex array, the
% _ArrayData_ array will include two columns
% (4 for sparse) to record the real and imaginary
% parts, and also "_ArrayIsComplex_":1 is added.
% opt.ParseLogical [0|1]: if this is set to 1, logical array elem
% will use true/false rather than 1/0.
% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single
% numerical element will be shown without a square
% bracket, unless it is the root object; if 0, square
% brackets are forced for any numerical arrays.
% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, savejson
% will use the name of the passed obj variable as the
% root object name; if obj is an expression and
% does not have a name, 'root' will be used; if this
% is set to 0 and rootname is empty, the root level
% will be merged down to the lower level.
% opt.Inf ['"$1_Inf_"'|string]: a customized regular expression pattern
% to represent +/-Inf. The matched pattern is '([-+]*)Inf'
% and $1 represents the sign. For those who want to use
% 1e999 to represent Inf, they can set opt.Inf to '$11e999'
% opt.NaN ['"_NaN_"'|string]: a customized regular expression pattern
% to represent NaN
% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),
% for example, if opt.JSONP='foo', the JSON data is
% wrapped inside a function call as 'foo(...);'
% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson
% back to the string form
% opt.SaveBinary [0|1]: 1 - save the JSON file in binary mode; 0 - text mode.
% opt.Compact [0|1]: 1- out compact JSON format (remove all newlines and tabs)
%
% opt can be replaced by a list of ('param',value) pairs. The param
% string is equivallent to a field in opt and is case sensitive.
% output:
% json: a string in the JSON format (see http://json.org)
%
% examples:
% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],...
% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...
% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...
% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...
% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...
% 'SpecialData',[nan, inf, -inf]);
% savejson('jmesh',jsonmesh)
% savejson('',jsonmesh,'ArrayIndent',0,'FloatFormat','\t%.5g')
%
% license:
% BSD, see LICENSE_BSD.txt files for details
%
% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)
%
if(nargin==1)
varname=inputname(1);
obj=rootname;
if(isempty(varname))
varname='root';
end
rootname=varname;
else
varname=inputname(2);
end
if(length(varargin)==1 && ischar(varargin{1}))
opt=struct('FileName',varargin{1});
else
opt=varargin2struct(varargin{:});
end
opt.IsOctave=exist('OCTAVE_VERSION','builtin');
rootisarray=0;
rootlevel=1;
forceroot=jsonopt('ForceRootName',0,opt);
if((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)
rootisarray=1;
rootlevel=0;
else
if(isempty(rootname))
rootname=varname;
end
end
if((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)
rootname='root';
end
whitespaces=struct('tab',sprintf('\t'),'newline',sprintf('\n'),'sep',sprintf(',\n'));
if(jsonopt('Compact',0,opt)==1)
whitespaces=struct('tab','','newline','','sep',',');
end
if(~isfield(opt,'whitespaces_'))
opt.whitespaces_=whitespaces;
end
nl=whitespaces.newline;
json=obj2json(rootname,obj,rootlevel,opt);
if(rootisarray)
json=sprintf('%s%s',json,nl);
else
json=sprintf('{%s%s%s}\n',nl,json,nl);
end
jsonp=jsonopt('JSONP','',opt);
if(~isempty(jsonp))
json=sprintf('%s(%s);%s',jsonp,json,nl);
end
% save to a file if FileName is set, suggested by Patrick Rapin
if(~isempty(jsonopt('FileName','',opt)))
if(jsonopt('SaveBinary',0,opt)==1)
fid = fopen(opt.FileName, 'wb');
fwrite(fid,json);
else
fid = fopen(opt.FileName, 'wt');
fwrite(fid,json,'char');
end
fclose(fid);
end
%%-------------------------------------------------------------------------
function txt=obj2json(name,item,level,varargin)
if(iscell(item))
txt=cell2json(name,item,level,varargin{:});
elseif(isstruct(item))
txt=struct2json(name,item,level,varargin{:});
elseif(ischar(item))
txt=str2json(name,item,level,varargin{:});
else
txt=mat2json(name,item,level,varargin{:});
end
%%-------------------------------------------------------------------------
function txt=cell2json(name,item,level,varargin)
txt='';
if(~iscell(item))
error('input is not a cell');
end
dim=size(item);
if(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now
item=reshape(item,dim(1),numel(item)/dim(1));
dim=size(item);
end
len=numel(item);
ws=jsonopt('whitespaces_',struct('tab',sprintf('\t'),'newline',sprintf('\n'),'sep',sprintf(',\n')),varargin{:});
padding0=repmat(ws.tab,1,level);
padding2=repmat(ws.tab,1,level+1);
nl=ws.newline;
if(len>1)
if(~isempty(name))
txt=sprintf('%s"%s": [%s',padding0, checkname(name,varargin{:}),nl); name='';
else
txt=sprintf('%s[%s',padding0,nl);
end
elseif(len==0)
if(~isempty(name))
txt=sprintf('%s"%s": []',padding0, checkname(name,varargin{:})); name='';
else
txt=sprintf('%s[]',padding0);
end
end
for j=1:dim(2)
if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end
for i=1:dim(1)
txt=sprintf('%s%s',txt,obj2json(name,item{i,j},level+(dim(1)>1)+1,varargin{:}));
if(i<dim(1)) txt=sprintf('%s%s',txt,sprintf(',%s',nl)); end
end
if(dim(1)>1) txt=sprintf('%s%s%s]',txt,nl,padding2); end
if(j<dim(2)) txt=sprintf('%s%s',txt,sprintf(',%s',nl)); end
%if(j==dim(2)) txt=sprintf('%s%s',txt,sprintf(',%s',nl)); end
end
if(len>1) txt=sprintf('%s%s%s]',txt,nl,padding0); end
%%-------------------------------------------------------------------------
function txt=struct2json(name,item,level,varargin)
txt='';
if(~isstruct(item))
error('input is not a struct');
end
dim=size(item);
if(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now
item=reshape(item,dim(1),numel(item)/dim(1));
dim=size(item);
end
len=numel(item);
ws=struct('tab',sprintf('\t'),'newline',sprintf('\n'));
ws=jsonopt('whitespaces_',ws,varargin{:});
padding0=repmat(ws.tab,1,level);
padding2=repmat(ws.tab,1,level+1);
padding1=repmat(ws.tab,1,level+(dim(1)>1)+(len>1));
nl=ws.newline;
if(~isempty(name))
if(len>1) txt=sprintf('%s"%s": [%s',padding0,checkname(name,varargin{:}),nl); end
else
if(len>1) txt=sprintf('%s[%s',padding0,nl); end
end
for j=1:dim(2)
if(dim(1)>1) txt=sprintf('%s%s[%s',txt,padding2,nl); end
for i=1:dim(1)
names = fieldnames(item(i,j));
if(~isempty(name) && len==1)
txt=sprintf('%s%s"%s": {%s',txt,padding1, checkname(name,varargin{:}),nl);
else
txt=sprintf('%s%s{%s',txt,padding1,nl);
end
if(~isempty(names))
for e=1:length(names)
txt=sprintf('%s%s',txt,obj2json(names{e},getfield(item(i,j),...
names{e}),level+(dim(1)>1)+1+(len>1),varargin{:}));
if(e<length(names)) txt=sprintf('%s%s',txt,','); end
txt=sprintf('%s%s',txt,nl);
end
end
txt=sprintf('%s%s}',txt,padding1);
if(i<dim(1)) txt=sprintf('%s%s',txt,sprintf(',%s',nl)); end
end
if(dim(1)>1) txt=sprintf('%s%s%s]',txt,nl,padding2); end
if(j<dim(2)) txt=sprintf('%s%s',txt,sprintf(',%s',nl)); end
end
if(len>1) txt=sprintf('%s%s%s]',txt,nl,padding0); end
%%-------------------------------------------------------------------------
function txt=str2json(name,item,level,varargin)
txt='';
if(~ischar(item))
error('input is not a string');
end
item=reshape(item, max(size(item),[1 0]));
len=size(item,1);
ws=struct('tab',sprintf('\t'),'newline',sprintf('\n'),'sep',sprintf(',\n'));
ws=jsonopt('whitespaces_',ws,varargin{:});
padding1=repmat(ws.tab,1,level);
padding0=repmat(ws.tab,1,level+1);
nl=ws.newline;
sep=ws.sep;
if(~isempty(name))
if(len>1) txt=sprintf('%s"%s": [%s',padding1,checkname(name,varargin{:}),nl); end
else
if(len>1) txt=sprintf('%s[%s',padding1,nl); end
end
isoct=jsonopt('IsOctave',0,varargin{:});
for e=1:len
if(isoct)
val=regexprep(item(e,:),'\\','\\');
val=regexprep(val,'"','\"');
val=regexprep(val,'^"','\"');
else
val=regexprep(item(e,:),'\\','\\\\');
val=regexprep(val,'"','\\"');
val=regexprep(val,'^"','\\"');
end
val=escapejsonstring(val);
if(len==1)
obj=['"' checkname(name,varargin{:}) '": ' '"',val,'"'];
if(isempty(name)) obj=['"',val,'"']; end
txt=sprintf('%s%s%s%s',txt,padding1,obj);
else
txt=sprintf('%s%s%s%s',txt,padding0,['"',val,'"']);
end
if(e==len) sep=''; end
txt=sprintf('%s%s',txt,sep);
end
if(len>1) txt=sprintf('%s%s%s%s',txt,nl,padding1,']'); end
%%-------------------------------------------------------------------------
function txt=mat2json(name,item,level,varargin)
if(~isnumeric(item) && ~islogical(item))
error('input is not an array');
end
ws=struct('tab',sprintf('\t'),'newline',sprintf('\n'),'sep',sprintf(',\n'));
ws=jsonopt('whitespaces_',ws,varargin{:});
padding1=repmat(ws.tab,1,level);
padding0=repmat(ws.tab,1,level+1);
nl=ws.newline;
sep=ws.sep;
if(length(size(item))>2 || issparse(item) || ~isreal(item) || ...
isempty(item) ||jsonopt('ArrayToStruct',0,varargin{:}))
if(isempty(name))
txt=sprintf('%s{%s%s"_ArrayType_": "%s",%s%s"_ArraySize_": %s,%s',...
padding1,nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\s+',','),nl);
else
txt=sprintf('%s"%s": {%s%s"_ArrayType_": "%s",%s%s"_ArraySize_": %s,%s',...
padding1,checkname(name,varargin{:}),nl,padding0,class(item),nl,padding0,regexprep(mat2str(size(item)),'\s+',','),nl);
end
else
if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1 && level>0)
numtxt=regexprep(regexprep(matdata2json(item,level+1,varargin{:}),'^\[',''),']','');
else
numtxt=matdata2json(item,level+1,varargin{:});
end
if(isempty(name))
txt=sprintf('%s%s',padding1,numtxt);
else
if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)
txt=sprintf('%s"%s": %s',padding1,checkname(name,varargin{:}),numtxt);
else
txt=sprintf('%s"%s": %s',padding1,checkname(name,varargin{:}),numtxt);
end
end
return;
end
dataformat='%s%s%s%s%s';
if(issparse(item))
[ix,iy]=find(item);
data=full(item(find(item)));
if(~isreal(item))
data=[real(data(:)),imag(data(:))];
if(size(item,1)==1)
% Kludge to have data's 'transposedness' match item's.
% (Necessary for complex row vector handling below.)
data=data';
end
txt=sprintf(dataformat,txt,padding0,'"_ArrayIsComplex_": ','1', sep);
end
txt=sprintf(dataformat,txt,padding0,'"_ArrayIsSparse_": ','1', sep);
if(size(item,1)==1)
% Row vector, store only column indices.
txt=sprintf(dataformat,txt,padding0,'"_ArrayData_": ',...
matdata2json([iy(:),data'],level+2,varargin{:}), nl);
elseif(size(item,2)==1)
% Column vector, store only row indices.
txt=sprintf(dataformat,txt,padding0,'"_ArrayData_": ',...
matdata2json([ix,data],level+2,varargin{:}), nl);
else
% General case, store row and column indices.
txt=sprintf(dataformat,txt,padding0,'"_ArrayData_": ',...
matdata2json([ix,iy,data],level+2,varargin{:}), nl);
end
else
if(isreal(item))
txt=sprintf(dataformat,txt,padding0,'"_ArrayData_": ',...
matdata2json(item(:)',level+2,varargin{:}), nl);
else
txt=sprintf(dataformat,txt,padding0,'"_ArrayIsComplex_": ','1', sep);
txt=sprintf(dataformat,txt,padding0,'"_ArrayData_": ',...
matdata2json([real(item(:)) imag(item(:))],level+2,varargin{:}), nl);
end
end
txt=sprintf('%s%s%s',txt,padding1,'}');
%%-------------------------------------------------------------------------
function txt=matdata2json(mat,level,varargin)
ws=struct('tab',sprintf('\t'),'newline',sprintf('\n'),'sep',sprintf(',\n'));
ws=jsonopt('whitespaces_',ws,varargin{:});
tab=ws.tab;
nl=ws.newline;
if(size(mat,1)==1)
pre='';
post='';
level=level-1;
else
pre=sprintf('[%s',nl);
post=sprintf('%s%s]',nl,repmat(tab,1,level-1));
end
if(isempty(mat))
txt='null';
return;
end
floatformat=jsonopt('FloatFormat','%.10g',varargin{:});
%if(numel(mat)>1)
formatstr=['[' repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf('],%s',nl)]];
%else
% formatstr=[repmat([floatformat ','],1,size(mat,2)-1) [floatformat sprintf(',\n')]];
%end
if(nargin>=2 && size(mat,1)>1 && jsonopt('ArrayIndent',1,varargin{:})==1)
formatstr=[repmat(tab,1,level) formatstr];
end
txt=sprintf(formatstr,mat');
txt(end-length(nl):end)=[];
if(islogical(mat) && jsonopt('ParseLogical',0,varargin{:})==1)
txt=regexprep(txt,'1','true');
txt=regexprep(txt,'0','false');
end
%txt=regexprep(mat2str(mat),'\s+',',');
%txt=regexprep(txt,';',sprintf('],\n['));
% if(nargin>=2 && size(mat,1)>1)
% txt=regexprep(txt,'\[',[repmat(sprintf('\t'),1,level) '[']);
% end
txt=[pre txt post];
if(any(isinf(mat(:))))
txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','"$1_Inf_"',varargin{:}));
end
if(any(isnan(mat(:))))
txt=regexprep(txt,'NaN',jsonopt('NaN','"_NaN_"',varargin{:}));
end
%%-------------------------------------------------------------------------
function newname=checkname(name,varargin)
isunpack=jsonopt('UnpackHex',1,varargin{:});
newname=name;
if(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))
return
end
if(isunpack)
isoct=jsonopt('IsOctave',0,varargin{:});
if(~isoct)
newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');
else
pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');
pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');
if(isempty(pos)) return; end
str0=name;
pos0=[0 pend(:)' length(name)];
newname='';
for i=1:length(pos)
newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];
end
if(pos(end)~=length(name))
newname=[newname str0(pos0(end-1)+1:pos0(end))];
end
end
end
%%-------------------------------------------------------------------------
function newstr=escapejsonstring(str)
newstr=str;
isoct=exist('OCTAVE_VERSION','builtin');
if(isoct)
vv=sscanf(OCTAVE_VERSION,'%f');
if(vv(1)>=3.8) isoct=0; end
end
if(isoct)
escapechars={'\a','\f','\n','\r','\t','\v'};
for i=1:length(escapechars);
newstr=regexprep(newstr,escapechars{i},escapechars{i});
end
else
escapechars={'\a','\b','\f','\n','\r','\t','\v'};
for i=1:length(escapechars);
newstr=regexprep(newstr,escapechars{i},regexprep(escapechars{i},'\\','\\\\'));
end
end

504
machine-learning-ex6/ex6/lib/jsonlab/saveubjson.m Normal file
View File

@@ -0,0 +1,504 @@
function json=saveubjson(rootname,obj,varargin)
%
% json=saveubjson(rootname,obj,filename)
% or
% json=saveubjson(rootname,obj,opt)
% json=saveubjson(rootname,obj,'param1',value1,'param2',value2,...)
%
% convert a MATLAB object (cell, struct or array) into a Universal
% Binary JSON (UBJSON) binary string
%
% author: Qianqian Fang (fangq<at> nmr.mgh.harvard.edu)
% created on 2013/08/17
%
% $Id: saveubjson.m 460 2015-01-03 00:30:45Z fangq $
%
% input:
% rootname: the name of the root-object, when set to '', the root name
% is ignored, however, when opt.ForceRootName is set to 1 (see below),
% the MATLAB variable name will be used as the root name.
% obj: a MATLAB object (array, cell, cell array, struct, struct array)
% filename: a string for the file name to save the output UBJSON data
% opt: a struct for additional options, ignore to use default values.
% opt can have the following fields (first in [.|.] is the default)
%
% opt.FileName [''|string]: a file name to save the output JSON data
% opt.ArrayToStruct[0|1]: when set to 0, saveubjson outputs 1D/2D
% array in JSON array format; if sets to 1, an
% array will be shown as a struct with fields
% "_ArrayType_", "_ArraySize_" and "_ArrayData_"; for
% sparse arrays, the non-zero elements will be
% saved to _ArrayData_ field in triplet-format i.e.
% (ix,iy,val) and "_ArrayIsSparse_" will be added
% with a value of 1; for a complex array, the
% _ArrayData_ array will include two columns
% (4 for sparse) to record the real and imaginary
% parts, and also "_ArrayIsComplex_":1 is added.
% opt.ParseLogical [1|0]: if this is set to 1, logical array elem
% will use true/false rather than 1/0.
% opt.NoRowBracket [1|0]: if this is set to 1, arrays with a single
% numerical element will be shown without a square
% bracket, unless it is the root object; if 0, square
% brackets are forced for any numerical arrays.
% opt.ForceRootName [0|1]: when set to 1 and rootname is empty, saveubjson
% will use the name of the passed obj variable as the
% root object name; if obj is an expression and
% does not have a name, 'root' will be used; if this
% is set to 0 and rootname is empty, the root level
% will be merged down to the lower level.
% opt.JSONP [''|string]: to generate a JSONP output (JSON with padding),
% for example, if opt.JSON='foo', the JSON data is
% wrapped inside a function call as 'foo(...);'
% opt.UnpackHex [1|0]: conver the 0x[hex code] output by loadjson
% back to the string form
%
% opt can be replaced by a list of ('param',value) pairs. The param
% string is equivallent to a field in opt and is case sensitive.
% output:
% json: a binary string in the UBJSON format (see http://ubjson.org)
%
% examples:
% jsonmesh=struct('MeshNode',[0 0 0;1 0 0;0 1 0;1 1 0;0 0 1;1 0 1;0 1 1;1 1 1],...
% 'MeshTetra',[1 2 4 8;1 3 4 8;1 2 6 8;1 5 6 8;1 5 7 8;1 3 7 8],...
% 'MeshTri',[1 2 4;1 2 6;1 3 4;1 3 7;1 5 6;1 5 7;...
% 2 8 4;2 8 6;3 8 4;3 8 7;5 8 6;5 8 7],...
% 'MeshCreator','FangQ','MeshTitle','T6 Cube',...
% 'SpecialData',[nan, inf, -inf]);
% saveubjson('jsonmesh',jsonmesh)
% saveubjson('jsonmesh',jsonmesh,'meshdata.ubj')
%
% license:
% BSD, see LICENSE_BSD.txt files for details
%
% -- this function is part of JSONLab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)
%
if(nargin==1)
varname=inputname(1);
obj=rootname;
if(isempty(varname))
varname='root';
end
rootname=varname;
else
varname=inputname(2);
end
if(length(varargin)==1 && ischar(varargin{1}))
opt=struct('FileName',varargin{1});
else
opt=varargin2struct(varargin{:});
end
opt.IsOctave=exist('OCTAVE_VERSION','builtin');
rootisarray=0;
rootlevel=1;
forceroot=jsonopt('ForceRootName',0,opt);
if((isnumeric(obj) || islogical(obj) || ischar(obj) || isstruct(obj) || iscell(obj)) && isempty(rootname) && forceroot==0)
rootisarray=1;
rootlevel=0;
else
if(isempty(rootname))
rootname=varname;
end
end
if((isstruct(obj) || iscell(obj))&& isempty(rootname) && forceroot)
rootname='root';
end
json=obj2ubjson(rootname,obj,rootlevel,opt);
if(~rootisarray)
json=['{' json '}'];
end
jsonp=jsonopt('JSONP','',opt);
if(~isempty(jsonp))
json=[jsonp '(' json ')'];
end
% save to a file if FileName is set, suggested by Patrick Rapin
if(~isempty(jsonopt('FileName','',opt)))
fid = fopen(opt.FileName, 'wb');
fwrite(fid,json);
fclose(fid);
end
%%-------------------------------------------------------------------------
function txt=obj2ubjson(name,item,level,varargin)
if(iscell(item))
txt=cell2ubjson(name,item,level,varargin{:});
elseif(isstruct(item))
txt=struct2ubjson(name,item,level,varargin{:});
elseif(ischar(item))
txt=str2ubjson(name,item,level,varargin{:});
else
txt=mat2ubjson(name,item,level,varargin{:});
end
%%-------------------------------------------------------------------------
function txt=cell2ubjson(name,item,level,varargin)
txt='';
if(~iscell(item))
error('input is not a cell');
end
dim=size(item);
if(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now
item=reshape(item,dim(1),numel(item)/dim(1));
dim=size(item);
end
len=numel(item); % let's handle 1D cell first
if(len>1)
if(~isempty(name))
txt=[S_(checkname(name,varargin{:})) '[']; name='';
else
txt='[';
end
elseif(len==0)
if(~isempty(name))
txt=[S_(checkname(name,varargin{:})) 'Z']; name='';
else
txt='Z';
end
end
for j=1:dim(2)
if(dim(1)>1) txt=[txt '[']; end
for i=1:dim(1)
txt=[txt obj2ubjson(name,item{i,j},level+(len>1),varargin{:})];
end
if(dim(1)>1) txt=[txt ']']; end
end
if(len>1) txt=[txt ']']; end
%%-------------------------------------------------------------------------
function txt=struct2ubjson(name,item,level,varargin)
txt='';
if(~isstruct(item))
error('input is not a struct');
end
dim=size(item);
if(ndims(squeeze(item))>2) % for 3D or higher dimensions, flatten to 2D for now
item=reshape(item,dim(1),numel(item)/dim(1));
dim=size(item);
end
len=numel(item);
if(~isempty(name))
if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end
else
if(len>1) txt='['; end
end
for j=1:dim(2)
if(dim(1)>1) txt=[txt '[']; end
for i=1:dim(1)
names = fieldnames(item(i,j));
if(~isempty(name) && len==1)
txt=[txt S_(checkname(name,varargin{:})) '{'];
else
txt=[txt '{'];
end
if(~isempty(names))
for e=1:length(names)
txt=[txt obj2ubjson(names{e},getfield(item(i,j),...
names{e}),level+(dim(1)>1)+1+(len>1),varargin{:})];
end
end
txt=[txt '}'];
end
if(dim(1)>1) txt=[txt ']']; end
end
if(len>1) txt=[txt ']']; end
%%-------------------------------------------------------------------------
function txt=str2ubjson(name,item,level,varargin)
txt='';
if(~ischar(item))
error('input is not a string');
end
item=reshape(item, max(size(item),[1 0]));
len=size(item,1);
if(~isempty(name))
if(len>1) txt=[S_(checkname(name,varargin{:})) '[']; end
else
if(len>1) txt='['; end
end
isoct=jsonopt('IsOctave',0,varargin{:});
for e=1:len
val=item(e,:);
if(len==1)
obj=['' S_(checkname(name,varargin{:})) '' '',S_(val),''];
if(isempty(name)) obj=['',S_(val),'']; end
txt=[txt,'',obj];
else
txt=[txt,'',['',S_(val),'']];
end
end
if(len>1) txt=[txt ']']; end
%%-------------------------------------------------------------------------
function txt=mat2ubjson(name,item,level,varargin)
if(~isnumeric(item) && ~islogical(item))
error('input is not an array');
end
if(length(size(item))>2 || issparse(item) || ~isreal(item) || ...
isempty(item) || jsonopt('ArrayToStruct',0,varargin{:}))
cid=I_(uint32(max(size(item))));
if(isempty(name))
txt=['{' S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1)) ];
else
if(isempty(item))
txt=[S_(checkname(name,varargin{:})),'Z'];
return;
else
txt=[S_(checkname(name,varargin{:})),'{',S_('_ArrayType_'),S_(class(item)),S_('_ArraySize_'),I_a(size(item),cid(1))];
end
end
else
if(isempty(name))
txt=matdata2ubjson(item,level+1,varargin{:});
else
if(numel(item)==1 && jsonopt('NoRowBracket',1,varargin{:})==1)
numtxt=regexprep(regexprep(matdata2ubjson(item,level+1,varargin{:}),'^\[',''),']','');
txt=[S_(checkname(name,varargin{:})) numtxt];
else
txt=[S_(checkname(name,varargin{:})),matdata2ubjson(item,level+1,varargin{:})];
end
end
return;
end
if(issparse(item))
[ix,iy]=find(item);
data=full(item(find(item)));
if(~isreal(item))
data=[real(data(:)),imag(data(:))];
if(size(item,1)==1)
% Kludge to have data's 'transposedness' match item's.
% (Necessary for complex row vector handling below.)
data=data';
end
txt=[txt,S_('_ArrayIsComplex_'),'T'];
end
txt=[txt,S_('_ArrayIsSparse_'),'T'];
if(size(item,1)==1)
% Row vector, store only column indices.
txt=[txt,S_('_ArrayData_'),...
matdata2ubjson([iy(:),data'],level+2,varargin{:})];
elseif(size(item,2)==1)
% Column vector, store only row indices.
txt=[txt,S_('_ArrayData_'),...
matdata2ubjson([ix,data],level+2,varargin{:})];
else
% General case, store row and column indices.
txt=[txt,S_('_ArrayData_'),...
matdata2ubjson([ix,iy,data],level+2,varargin{:})];
end
else
if(isreal(item))
txt=[txt,S_('_ArrayData_'),...
matdata2ubjson(item(:)',level+2,varargin{:})];
else
txt=[txt,S_('_ArrayIsComplex_'),'T'];
txt=[txt,S_('_ArrayData_'),...
matdata2ubjson([real(item(:)) imag(item(:))],level+2,varargin{:})];
end
end
txt=[txt,'}'];
%%-------------------------------------------------------------------------
function txt=matdata2ubjson(mat,level,varargin)
if(isempty(mat))
txt='Z';
return;
end
if(size(mat,1)==1)
level=level-1;
end
type='';
hasnegtive=(mat<0);
if(isa(mat,'integer') || isinteger(mat) || (isfloat(mat) && all(mod(mat(:),1) == 0)))
if(isempty(hasnegtive))
if(max(mat(:))<=2^8)
type='U';
end
end
if(isempty(type))
% todo - need to consider negative ones separately
id= histc(abs(max(mat(:))),[0 2^7 2^15 2^31 2^63]);
if(isempty(find(id)))
error('high-precision data is not yet supported');
end
key='iIlL';
type=key(find(id));
end
txt=[I_a(mat(:),type,size(mat))];
elseif(islogical(mat))
logicalval='FT';
if(numel(mat)==1)
txt=logicalval(mat+1);
else
txt=['[$U#' I_a(size(mat),'l') typecast(swapbytes(uint8(mat(:)')),'uint8')];
end
else
if(numel(mat)==1)
txt=['[' D_(mat) ']'];
else
txt=D_a(mat(:),'D',size(mat));
end
end
%txt=regexprep(mat2str(mat),'\s+',',');
%txt=regexprep(txt,';',sprintf('],['));
% if(nargin>=2 && size(mat,1)>1)
% txt=regexprep(txt,'\[',[repmat(sprintf('\t'),1,level) '[']);
% end
if(any(isinf(mat(:))))
txt=regexprep(txt,'([-+]*)Inf',jsonopt('Inf','"$1_Inf_"',varargin{:}));
end
if(any(isnan(mat(:))))
txt=regexprep(txt,'NaN',jsonopt('NaN','"_NaN_"',varargin{:}));
end
%%-------------------------------------------------------------------------
function newname=checkname(name,varargin)
isunpack=jsonopt('UnpackHex',1,varargin{:});
newname=name;
if(isempty(regexp(name,'0x([0-9a-fA-F]+)_','once')))
return
end
if(isunpack)
isoct=jsonopt('IsOctave',0,varargin{:});
if(~isoct)
newname=regexprep(name,'(^x|_){1}0x([0-9a-fA-F]+)_','${native2unicode(hex2dec($2))}');
else
pos=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','start');
pend=regexp(name,'(^x|_){1}0x([0-9a-fA-F]+)_','end');
if(isempty(pos)) return; end
str0=name;
pos0=[0 pend(:)' length(name)];
newname='';
for i=1:length(pos)
newname=[newname str0(pos0(i)+1:pos(i)-1) char(hex2dec(str0(pos(i)+3:pend(i)-1)))];
end
if(pos(end)~=length(name))
newname=[newname str0(pos0(end-1)+1:pos0(end))];
end
end
end
%%-------------------------------------------------------------------------
function val=S_(str)
if(length(str)==1)
val=['C' str];
else
val=['S' I_(int32(length(str))) str];
end
%%-------------------------------------------------------------------------
function val=I_(num)
if(~isinteger(num))
error('input is not an integer');
end
if(num>=0 && num<255)
val=['U' data2byte(swapbytes(cast(num,'uint8')),'uint8')];
return;
end
key='iIlL';
cid={'int8','int16','int32','int64'};
for i=1:4
if((num>0 && num<2^(i*8-1)) || (num<0 && num>=-2^(i*8-1)))
val=[key(i) data2byte(swapbytes(cast(num,cid{i})),'uint8')];
return;
end
end
error('unsupported integer');
%%-------------------------------------------------------------------------
function val=D_(num)
if(~isfloat(num))
error('input is not a float');
end
if(isa(num,'single'))
val=['d' data2byte(num,'uint8')];
else
val=['D' data2byte(num,'uint8')];
end
%%-------------------------------------------------------------------------
function data=I_a(num,type,dim,format)
id=find(ismember('iUIlL',type));
if(id==0)
error('unsupported integer array');
end
% based on UBJSON specs, all integer types are stored in big endian format
if(id==1)
data=data2byte(swapbytes(int8(num)),'uint8');
blen=1;
elseif(id==2)
data=data2byte(swapbytes(uint8(num)),'uint8');
blen=1;
elseif(id==3)
data=data2byte(swapbytes(int16(num)),'uint8');
blen=2;
elseif(id==4)
data=data2byte(swapbytes(int32(num)),'uint8');
blen=4;
elseif(id==5)
data=data2byte(swapbytes(int64(num)),'uint8');
blen=8;
end
if(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))
format='opt';
end
if((nargin<4 || strcmp(format,'opt')) && numel(num)>1)
if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))
cid=I_(uint32(max(dim)));
data=['$' type '#' I_a(dim,cid(1)) data(:)'];
else
data=['$' type '#' I_(int32(numel(data)/blen)) data(:)'];
end
data=['[' data(:)'];
else
data=reshape(data,blen,numel(data)/blen);
data(2:blen+1,:)=data;
data(1,:)=type;
data=data(:)';
data=['[' data(:)' ']'];
end
%%-------------------------------------------------------------------------
function data=D_a(num,type,dim,format)
id=find(ismember('dD',type));
if(id==0)
error('unsupported float array');
end
if(id==1)
data=data2byte(single(num),'uint8');
elseif(id==2)
data=data2byte(double(num),'uint8');
end
if(nargin>=3 && length(dim)>=2 && prod(dim)~=dim(2))
format='opt';
end
if((nargin<4 || strcmp(format,'opt')) && numel(num)>1)
if(nargin>=3 && (length(dim)==1 || (length(dim)>=2 && prod(dim)~=dim(2))))
cid=I_(uint32(max(dim)));
data=['$' type '#' I_a(dim,cid(1)) data(:)'];
else
data=['$' type '#' I_(int32(numel(data)/(id*4))) data(:)'];
end
data=['[' data];
else
data=reshape(data,(id*4),length(data)/(id*4));
data(2:(id*4+1),:)=data;
data(1,:)=type;
data=data(:)';
data=['[' data(:)' ']'];
end
%%-------------------------------------------------------------------------
function bytes=data2byte(varargin)
bytes=typecast(varargin{:});
bytes=bytes(:)';

40
machine-learning-ex6/ex6/lib/jsonlab/varargin2struct.m Normal file
View File

@@ -0,0 +1,40 @@
function opt=varargin2struct(varargin)
%
% opt=varargin2struct('param1',value1,'param2',value2,...)
% or
% opt=varargin2struct(...,optstruct,...)
%
% convert a series of input parameters into a structure
%
% authors:Qianqian Fang (fangq<at> nmr.mgh.harvard.edu)
% date: 2012/12/22
%
% input:
% 'param', value: the input parameters should be pairs of a string and a value
% optstruct: if a parameter is a struct, the fields will be merged to the output struct
%
% output:
% opt: a struct where opt.param1=value1, opt.param2=value2 ...
%
% license:
% BSD, see LICENSE_BSD.txt files for details
%
% -- this function is part of jsonlab toolbox (http://iso2mesh.sf.net/cgi-bin/index.cgi?jsonlab)
%
len=length(varargin);
opt=struct;
if(len==0) return; end
i=1;
while(i<=len)
if(isstruct(varargin{i}))
opt=mergestruct(opt,varargin{i});
elseif(ischar(varargin{i}) && i<len)
opt=setfield(opt,varargin{i},varargin{i+1});
i=i+1;
else
error('input must be in the form of ...,''name'',value,... pairs or structs');
end
i=i+1;
end

30
machine-learning-ex6/ex6/lib/makeValidFieldName.m Normal file
View File

@@ -0,0 +1,30 @@
function str = makeValidFieldName(str)
% From MATLAB doc: field names must begin with a letter, which may be
% followed by any combination of letters, digits, and underscores.
% Invalid characters will be converted to underscores, and the prefix
% "x0x[Hex code]_" will be added if the first character is not a letter.
isoct=exist('OCTAVE_VERSION','builtin');
pos=regexp(str,'^[^A-Za-z]','once');
if(~isempty(pos))
if(~isoct)
str=regexprep(str,'^([^A-Za-z])','x0x${sprintf(''%X'',unicode2native($1))}_','once');
else
str=sprintf('x0x%X_%s',char(str(1)),str(2:end));
end
end
if(isempty(regexp(str,'[^0-9A-Za-z_]', 'once' ))) return; end
if(~isoct)
str=regexprep(str,'([^0-9A-Za-z_])','_0x${sprintf(''%X'',unicode2native($1))}_');
else
pos=regexp(str,'[^0-9A-Za-z_]');
if(isempty(pos)) return; end
str0=str;
pos0=[0 pos(:)' length(str)];
str='';
for i=1:length(pos)
str=[str str0(pos0(i)+1:pos(i)-1) sprintf('_0x%X_',str0(pos(i)))];
end
if(pos(end)~=length(str))
str=[str str0(pos0(end-1)+1:pos0(end))];
end
end

125
machine-learning-ex6/ex6/lib/submitWithConfiguration.m Normal file
View File

@@ -0,0 +1,125 @@
function submitWithConfiguration(conf)
addpath('./lib/jsonlab');
parts = parts(conf);
fprintf('== Submitting solutions | %s...\n', conf.itemName);
tokenFile = 'token.mat';
if exist(tokenFile, 'file')
load(tokenFile);
[email token] = promptToken(email, token, tokenFile);
else
[email token] = promptToken('', '', tokenFile);
end
if isempty(token)
fprintf('!! Submission Cancelled\n');
return
end
try
response = submitParts(conf, email, token, parts);
catch
e = lasterror();
fprintf( ...
'!! Submission failed: unexpected error: %s\n', ...
e.message);
fprintf('!! Please try again later.\n');
return
end
if isfield(response, 'errorMessage')
fprintf('!! Submission failed: %s\n', response.errorMessage);
else
showFeedback(parts, response);
save(tokenFile, 'email', 'token');
end
end
function [email token] = promptToken(email, existingToken, tokenFile)
if (~isempty(email) && ~isempty(existingToken))
prompt = sprintf( ...
'Use token from last successful submission (%s)? (Y/n): ', ...
email);
reenter = input(prompt, 's');
if (isempty(reenter) || reenter(1) == 'Y' || reenter(1) == 'y')
token = existingToken;
return;
else
delete(tokenFile);
end
end
email = input('Login (email address): ', 's');
token = input('Token: ', 's');
end
function isValid = isValidPartOptionIndex(partOptions, i)
isValid = (~isempty(i)) && (1 <= i) && (i <= numel(partOptions));
end
function response = submitParts(conf, email, token, parts)
body = makePostBody(conf, email, token, parts);
submissionUrl = submissionUrl();
params = {'jsonBody', body};
responseBody = urlread(submissionUrl, 'post', params);
response = loadjson(responseBody);
end
function body = makePostBody(conf, email, token, parts)
bodyStruct.assignmentSlug = conf.assignmentSlug;
bodyStruct.submitterEmail = email;
bodyStruct.secret = token;
bodyStruct.parts = makePartsStruct(conf, parts);
opt.Compact = 1;
body = savejson('', bodyStruct, opt);
end
function partsStruct = makePartsStruct(conf, parts)
for part = parts
partId = part{:}.id;
fieldName = makeValidFieldName(partId);
outputStruct.output = conf.output(partId);
partsStruct.(fieldName) = outputStruct;
end
end
function [parts] = parts(conf)
parts = {};
for partArray = conf.partArrays
part.id = partArray{:}{1};
part.sourceFiles = partArray{:}{2};
part.name = partArray{:}{3};
parts{end + 1} = part;
end
end
function showFeedback(parts, response)
fprintf('== \n');
fprintf('== %43s | %9s | %-s\n', 'Part Name', 'Score', 'Feedback');
fprintf('== %43s | %9s | %-s\n', '---------', '-----', '--------');
for part = parts
score = '';
partFeedback = '';
partFeedback = response.partFeedbacks.(makeValidFieldName(part{:}.id));
partEvaluation = response.partEvaluations.(makeValidFieldName(part{:}.id));
score = sprintf('%d / %3d', partEvaluation.score, partEvaluation.maxScore);
fprintf('== %43s | %9s | %-s\n', part{:}.name, score, partFeedback);
end
evaluation = response.evaluation;
totalScore = sprintf('%d / %d', evaluation.score, evaluation.maxScore);
fprintf('== --------------------------------\n');
fprintf('== %43s | %9s | %-s\n', '', totalScore, '');
fprintf('== \n');
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
% Service configuration
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function submissionUrl = submissionUrl()
submissionUrl = 'https://www-origin.coursera.org/api/onDemandProgrammingImmediateFormSubmissions.v1';
end

12
machine-learning-ex6/ex6/linearKernel.m Normal file
View File

@@ -0,0 +1,12 @@
function sim = linearKernel(x1, x2)
%LINEARKERNEL returns a linear kernel between x1 and x2
% sim = linearKernel(x1, x2) returns a linear kernel between x1 and x2
% and returns the value in sim
% Ensure that x1 and x2 are column vectors
x1 = x1(:); x2 = x2(:);
% Compute the kernel
sim = x1' * x2; % dot product
end

17
machine-learning-ex6/ex6/plotData.m Normal file
View File

@@ -0,0 +1,17 @@
function plotData(X, y)
%PLOTDATA Plots the data points X and y into a new figure
% PLOTDATA(x,y) plots the data points with + for the positive examples
% and o for the negative examples. X is assumed to be a Mx2 matrix.
%
% Note: This was slightly modified such that it expects y = 1 or y = 0
% Find Indices of Positive and Negative Examples
pos = find(y == 1); neg = find(y == 0);
% Plot Examples
plot(X(pos, 1), X(pos, 2), 'k+','LineWidth', 1, 'MarkerSize', 7)
hold on;
plot(X(neg, 1), X(neg, 2), 'ko', 'MarkerFaceColor', 'y', 'MarkerSize', 7)
hold off;
end

385
machine-learning-ex6/ex6/porterStemmer.m Normal file
View File

@@ -0,0 +1,385 @@
function stem = porterStemmer(inString)
% Applies the Porter Stemming algorithm as presented in the following
% paper:
% Porter, 1980, An algorithm for suffix stripping, Program, Vol. 14,
% no. 3, pp 130-137
% Original code modeled after the C version provided at:
% http://www.tartarus.org/~martin/PorterStemmer/c.txt
% The main part of the stemming algorithm starts here. b is an array of
% characters, holding the word to be stemmed. The letters are in b[k0],
% b[k0+1] ending at b[k]. In fact k0 = 1 in this demo program (since
% matlab begins indexing by 1 instead of 0). k is readjusted downwards as
% the stemming progresses. Zero termination is not in fact used in the
% algorithm.
% To call this function, use the string to be stemmed as the input
% argument. This function returns the stemmed word as a string.
% Lower-case string
inString = lower(inString);
global j;
b = inString;
k = length(b);
k0 = 1;
j = k;
% With this if statement, strings of length 1 or 2 don't go through the
% stemming process. Remove this conditional to match the published
% algorithm.
stem = b;
if k > 2
% Output displays per step are commented out.
%disp(sprintf('Word to stem: %s', b));
x = step1ab(b, k, k0);
%disp(sprintf('Steps 1A and B yield: %s', x{1}));
x = step1c(x{1}, x{2}, k0);
%disp(sprintf('Step 1C yields: %s', x{1}));
x = step2(x{1}, x{2}, k0);
%disp(sprintf('Step 2 yields: %s', x{1}));
x = step3(x{1}, x{2}, k0);
%disp(sprintf('Step 3 yields: %s', x{1}));
x = step4(x{1}, x{2}, k0);
%disp(sprintf('Step 4 yields: %s', x{1}));
x = step5(x{1}, x{2}, k0);
%disp(sprintf('Step 5 yields: %s', x{1}));
stem = x{1};
end
% cons(j) is TRUE <=> b[j] is a consonant.
function c = cons(i, b, k0)
c = true;
switch(b(i))
case {'a', 'e', 'i', 'o', 'u'}
c = false;
case 'y'
if i == k0
c = true;
else
c = ~cons(i - 1, b, k0);
end
end
% mseq() measures the number of consonant sequences between k0 and j. If
% c is a consonant sequence and v a vowel sequence, and <..> indicates
% arbitrary presence,
% <c><v> gives 0
% <c>vc<v> gives 1
% <c>vcvc<v> gives 2
% <c>vcvcvc<v> gives 3
% ....
function n = measure(b, k0)
global j;
n = 0;
i = k0;
while true
if i > j
return
end
if ~cons(i, b, k0)
break;
end
i = i + 1;
end
i = i + 1;
while true
while true
if i > j
return
end
if cons(i, b, k0)
break;
end
i = i + 1;
end
i = i + 1;
n = n + 1;
while true
if i > j
return
end
if ~cons(i, b, k0)
break;
end
i = i + 1;
end
i = i + 1;
end
% vowelinstem() is TRUE <=> k0,...j contains a vowel
function vis = vowelinstem(b, k0)
global j;
for i = k0:j,
if ~cons(i, b, k0)
vis = true;
return
end
end
vis = false;
%doublec(i) is TRUE <=> i,(i-1) contain a double consonant.
function dc = doublec(i, b, k0)
if i < k0+1
dc = false;
return
end
if b(i) ~= b(i-1)
dc = false;
return
end
dc = cons(i, b, k0);
% cvc(j) is TRUE <=> j-2,j-1,j has the form consonant - vowel - consonant
% and also if the second c is not w,x or y. this is used when trying to
% restore an e at the end of a short word. e.g.
%
% cav(e), lov(e), hop(e), crim(e), but
% snow, box, tray.
function c1 = cvc(i, b, k0)
if ((i < (k0+2)) || ~cons(i, b, k0) || cons(i-1, b, k0) || ~cons(i-2, b, k0))
c1 = false;
else
if (b(i) == 'w' || b(i) == 'x' || b(i) == 'y')
c1 = false;
return
end
c1 = true;
end
% ends(s) is TRUE <=> k0,...k ends with the string s.
function s = ends(str, b, k)
global j;
if (str(length(str)) ~= b(k))
s = false;
return
end % tiny speed-up
if (length(str) > k)
s = false;
return
end
if strcmp(b(k-length(str)+1:k), str)
s = true;
j = k - length(str);
return
else
s = false;
end
% setto(s) sets (j+1),...k to the characters in the string s, readjusting
% k accordingly.
function so = setto(s, b, k)
global j;
for i = j+1:(j+length(s))
b(i) = s(i-j);
end
if k > j+length(s)
b((j+length(s)+1):k) = '';
end
k = length(b);
so = {b, k};
% rs(s) is used further down.
% [Note: possible null/value for r if rs is called]
function r = rs(str, b, k, k0)
r = {b, k};
if measure(b, k0) > 0
r = setto(str, b, k);
end
% step1ab() gets rid of plurals and -ed or -ing. e.g.
% caresses -> caress
% ponies -> poni
% ties -> ti
% caress -> caress
% cats -> cat
% feed -> feed
% agreed -> agree
% disabled -> disable
% matting -> mat
% mating -> mate
% meeting -> meet
% milling -> mill
% messing -> mess
% meetings -> meet
function s1ab = step1ab(b, k, k0)
global j;
if b(k) == 's'
if ends('sses', b, k)
k = k-2;
elseif ends('ies', b, k)
retVal = setto('i', b, k);
b = retVal{1};
k = retVal{2};
elseif (b(k-1) ~= 's')
k = k-1;
end
end
if ends('eed', b, k)
if measure(b, k0) > 0;
k = k-1;
end
elseif (ends('ed', b, k) || ends('ing', b, k)) && vowelinstem(b, k0)
k = j;
retVal = {b, k};
if ends('at', b, k)
retVal = setto('ate', b(k0:k), k);
elseif ends('bl', b, k)
retVal = setto('ble', b(k0:k), k);
elseif ends('iz', b, k)
retVal = setto('ize', b(k0:k), k);
elseif doublec(k, b, k0)
retVal = {b, k-1};
if b(retVal{2}) == 'l' || b(retVal{2}) == 's' || ...
b(retVal{2}) == 'z'
retVal = {retVal{1}, retVal{2}+1};
end
elseif measure(b, k0) == 1 && cvc(k, b, k0)
retVal = setto('e', b(k0:k), k);
end
k = retVal{2};
b = retVal{1}(k0:k);
end
j = k;
s1ab = {b(k0:k), k};
% step1c() turns terminal y to i when there is another vowel in the stem.
function s1c = step1c(b, k, k0)
global j;
if ends('y', b, k) && vowelinstem(b, k0)
b(k) = 'i';
end
j = k;
s1c = {b, k};
% step2() maps double suffices to single ones. so -ization ( = -ize plus
% -ation) maps to -ize etc. note that the string before the suffix must give
% m() > 0.
function s2 = step2(b, k, k0)
global j;
s2 = {b, k};
switch b(k-1)
case {'a'}
if ends('ational', b, k) s2 = rs('ate', b, k, k0);
elseif ends('tional', b, k) s2 = rs('tion', b, k, k0); end;
case {'c'}
if ends('enci', b, k) s2 = rs('ence', b, k, k0);
elseif ends('anci', b, k) s2 = rs('ance', b, k, k0); end;
case {'e'}
if ends('izer', b, k) s2 = rs('ize', b, k, k0); end;
case {'l'}
if ends('bli', b, k) s2 = rs('ble', b, k, k0);
elseif ends('alli', b, k) s2 = rs('al', b, k, k0);
elseif ends('entli', b, k) s2 = rs('ent', b, k, k0);
elseif ends('eli', b, k) s2 = rs('e', b, k, k0);
elseif ends('ousli', b, k) s2 = rs('ous', b, k, k0); end;
case {'o'}
if ends('ization', b, k) s2 = rs('ize', b, k, k0);
elseif ends('ation', b, k) s2 = rs('ate', b, k, k0);
elseif ends('ator', b, k) s2 = rs('ate', b, k, k0); end;
case {'s'}
if ends('alism', b, k) s2 = rs('al', b, k, k0);
elseif ends('iveness', b, k) s2 = rs('ive', b, k, k0);
elseif ends('fulness', b, k) s2 = rs('ful', b, k, k0);
elseif ends('ousness', b, k) s2 = rs('ous', b, k, k0); end;
case {'t'}
if ends('aliti', b, k) s2 = rs('al', b, k, k0);
elseif ends('iviti', b, k) s2 = rs('ive', b, k, k0);
elseif ends('biliti', b, k) s2 = rs('ble', b, k, k0); end;
case {'g'}
if ends('logi', b, k) s2 = rs('log', b, k, k0); end;
end
j = s2{2};
% step3() deals with -ic-, -full, -ness etc. similar strategy to step2.
function s3 = step3(b, k, k0)
global j;
s3 = {b, k};
switch b(k)
case {'e'}
if ends('icate', b, k) s3 = rs('ic', b, k, k0);
elseif ends('ative', b, k) s3 = rs('', b, k, k0);
elseif ends('alize', b, k) s3 = rs('al', b, k, k0); end;
case {'i'}
if ends('iciti', b, k) s3 = rs('ic', b, k, k0); end;
case {'l'}
if ends('ical', b, k) s3 = rs('ic', b, k, k0);
elseif ends('ful', b, k) s3 = rs('', b, k, k0); end;
case {'s'}
if ends('ness', b, k) s3 = rs('', b, k, k0); end;
end
j = s3{2};
% step4() takes off -ant, -ence etc., in context <c>vcvc<v>.
function s4 = step4(b, k, k0)
global j;
switch b(k-1)
case {'a'}
if ends('al', b, k) end;
case {'c'}
if ends('ance', b, k)
elseif ends('ence', b, k) end;
case {'e'}
if ends('er', b, k) end;
case {'i'}
if ends('ic', b, k) end;
case {'l'}
if ends('able', b, k)
elseif ends('ible', b, k) end;
case {'n'}
if ends('ant', b, k)
elseif ends('ement', b, k)
elseif ends('ment', b, k)
elseif ends('ent', b, k) end;
case {'o'}
if ends('ion', b, k)
if j == 0
elseif ~(strcmp(b(j),'s') || strcmp(b(j),'t'))
j = k;
end
elseif ends('ou', b, k) end;
case {'s'}
if ends('ism', b, k) end;
case {'t'}
if ends('ate', b, k)
elseif ends('iti', b, k) end;
case {'u'}
if ends('ous', b, k) end;
case {'v'}
if ends('ive', b, k) end;
case {'z'}
if ends('ize', b, k) end;
end
if measure(b, k0) > 1
s4 = {b(k0:j), j};
else
s4 = {b(k0:k), k};
end
% step5() removes a final -e if m() > 1, and changes -ll to -l if m() > 1.
function s5 = step5(b, k, k0)
global j;
j = k;
if b(k) == 'e'
a = measure(b, k0);
if (a > 1) || ((a == 1) && ~cvc(k-1, b, k0))
k = k-1;
end
end
if (b(k) == 'l') && doublec(k, b, k0) && (measure(b, k0) > 1)
k = k-1;
end
s5 = {b(k0:k), k};

125
machine-learning-ex6/ex6/processEmail.m Normal file
View File

@@ -0,0 +1,125 @@
function word_indices = processEmail(email_contents)
%PROCESSEMAIL preprocesses a the body of an email and
%returns a list of word_indices
% word_indices = PROCESSEMAIL(email_contents) preprocesses
% the body of an email and returns a list of indices of the
% words contained in the email.
%
% Load Vocabulary
vocabList = getVocabList();
% Init return value
word_indices = [];
% ========================== Preprocess Email ===========================
% Find the Headers ( \n\n and remove )
% Uncomment the following lines if you are working with raw emails with the
% full headers
% hdrstart = strfind(email_contents, ([char(10) char(10)]));
% email_contents = email_contents(hdrstart(1):end);
% Lower case
email_contents = lower(email_contents);
% Strip all HTML
% Looks for any expression that starts with < and ends with > and replace
% and does not have any < or > in the tag it with a space
email_contents = regexprep(email_contents, '<[^<>]+>', ' ');
% Handle Numbers
% Look for one or more characters between 0-9
email_contents = regexprep(email_contents, '[0-9]+', 'number');
% Handle URLS
% Look for strings starting with http:// or https://
email_contents = regexprep(email_contents, ...
'(http|https)://[^\s]*', 'httpaddr');
% Handle Email Addresses
% Look for strings with @ in the middle
email_contents = regexprep(email_contents, '[^\s]+@[^\s]+', 'emailaddr');
% Handle $ sign
email_contents = regexprep(email_contents, '[$]+', 'dollar');
% ========================== Tokenize Email ===========================
% Output the email to screen as well
fprintf('\n==== Processed Email ====\n\n');
% Process file
l = 0;
while ~isempty(email_contents)
% Tokenize and also get rid of any punctuation
[str, email_contents] = ...
strtok(email_contents, ...
[' @$/#.-:&*+=[]?!(){},''">_<;%' char(10) char(13)]);
% Remove any non alphanumeric characters
str = regexprep(str, '[^a-zA-Z0-9]', '');
% Stem the word
% (the porterStemmer sometimes has issues, so we use a try catch block)
try str = porterStemmer(strtrim(str));
catch str = ''; continue;
end;
% Skip the word if it is too short
if length(str) < 1
continue;
end
% Look up the word in the dictionary and add to word_indices if
% found
% ====================== YOUR CODE HERE ======================
% Instructions: Fill in this function to add the index of str to
% word_indices if it is in the vocabulary. At this point
% of the code, you have a stemmed word from the email in
% the variable str. You should look up str in the
% vocabulary list (vocabList). If a match exists, you
% should add the index of the word to the word_indices
% vector. Concretely, if str = 'action', then you should
% look up the vocabulary list to find where in vocabList
% 'action' appears. For example, if vocabList{18} =
% 'action', then, you should add 18 to the word_indices
% vector (e.g., word_indices = [word_indices ; 18]; ).
%
% Note: vocabList{idx} returns a the word with index idx in the
% vocabulary list.
%
% Note: You can use strcmp(str1, str2) to compare two strings (str1 and
% str2). It will return 1 only if the two strings are equivalent.
%
% =============================================================
% Print to screen, ensuring that the output lines are not too long
if (l + length(str) + 1) > 78
fprintf('\n');
l = 0;
end
fprintf('%s ', str);
l = l + length(str) + 1;
end
% Print footer
fprintf('\n\n=========================\n');
end

18
machine-learning-ex6/ex6/readFile.m Normal file
View File

@@ -0,0 +1,18 @@
function file_contents = readFile(filename)
%READFILE reads a file and returns its entire contents
% file_contents = READFILE(filename) reads a file and returns its entire
% contents in file_contents
%
% Load File
fid = fopen(filename);
if fid
file_contents = fscanf(fid, '%c', inf);
fclose(fid);
else
file_contents = '';
fprintf('Unable to open %s\n', filename);
end
end

42
machine-learning-ex6/ex6/spamSample1.txt Normal file
View File

@@ -0,0 +1,42 @@
Do You Want To Make $1000 Or More Per Week?
If you are a motivated and qualified individual - I
will personally demonstrate to you a system that will
make you $1,000 per week or more! This is NOT mlm.
Call our 24 hour pre-recorded number to get the
details.
000-456-789
I need people who want to make serious money. Make
the call and get the facts.
Invest 2 minutes in yourself now!
000-456-789
Looking forward to your call and I will introduce you
to people like yourself who
are currently making $10,000 plus per week!
000-456-789
3484lJGv6-241lEaN9080lRmS6-271WxHo7524qiyT5-438rjUv5615hQcf0-662eiDB9057dMtVl72

8
machine-learning-ex6/ex6/spamSample2.txt Normal file
View File

@@ -0,0 +1,8 @@
Best Buy Viagra Generic Online
Viagra 100mg x 60 Pills $125, Free Pills & Reorder Discount, Top Selling 100% Quality & Satisfaction guaranteed!
We accept VISA, Master & E-Check Payments, 90000+ Satisfied Customers!
http://medphysitcstech.ru

BIN
machine-learning-ex6/ex6/spamTest.mat Normal file
View File

Binary file not shown.

BIN
machine-learning-ex6/ex6/spamTrain.mat Normal file
View File

Binary file not shown.

55
machine-learning-ex6/ex6/submit.m Normal file
View File

@@ -0,0 +1,55 @@
function submit()
addpath('./lib');
conf.assignmentSlug = 'support-vector-machines';
conf.itemName = 'Support Vector Machines';
conf.partArrays = { ...
{ ...
'1', ...
{ 'gaussianKernel.m' }, ...
'Gaussian Kernel', ...
}, ...
{ ...
'2', ...
{ 'dataset3Params.m' }, ...
'Parameters (C, sigma) for Dataset 3', ...
}, ...
{ ...
'3', ...
{ 'processEmail.m' }, ...
'Email Preprocessing', ...
}, ...
{ ...
'4', ...
{ 'emailFeatures.m' }, ...
'Email Feature Extraction', ...
}, ...
};
conf.output = @output;
submitWithConfiguration(conf);
end
function out = output(partId, auxstring)
% Random Test Cases
x1 = sin(1:10)';
x2 = cos(1:10)';
ec = 'the quick brown fox jumped over the lazy dog';
wi = 1 + abs(round(x1 * 1863));
wi = [wi ; wi];
if partId == '1'
sim = gaussianKernel(x1, x2, 2);
out = sprintf('%0.5f ', sim);
elseif partId == '2'
load('ex6data3.mat');
[C, sigma] = dataset3Params(X, y, Xval, yval);
out = sprintf('%0.5f ', C);
out = [out sprintf('%0.5f ', sigma)];
elseif partId == '3'
word_indices = processEmail(ec);
out = sprintf('%d ', word_indices);
elseif partId == '4'
x = emailFeatures(wi);
out = sprintf('%d ', x);
end
end

54
machine-learning-ex6/ex6/svmPredict.m Normal file
View File

@@ -0,0 +1,54 @@
function pred = svmPredict(model, X)
%SVMPREDICT returns a vector of predictions using a trained SVM model
%(svmTrain).
% pred = SVMPREDICT(model, X) returns a vector of predictions using a
% trained SVM model (svmTrain). X is a mxn matrix where there each
% example is a row. model is a svm model returned from svmTrain.
% predictions pred is a m x 1 column of predictions of {0, 1} values.
%
% Check if we are getting a column vector, if so, then assume that we only
% need to do prediction for a single example
if (size(X, 2) == 1)
% Examples should be in rows
X = X';
end
% Dataset
m = size(X, 1);
p = zeros(m, 1);
pred = zeros(m, 1);
if strcmp(func2str(model.kernelFunction), 'linearKernel')
% We can use the weights and bias directly if working with the
% linear kernel
p = X * model.w + model.b;
elseif strfind(func2str(model.kernelFunction), 'gaussianKernel')
% Vectorized RBF Kernel
% This is equivalent to computing the kernel on every pair of examples
X1 = sum(X.^2, 2);
X2 = sum(model.X.^2, 2)';
K = bsxfun(@plus, X1, bsxfun(@plus, X2, - 2 * X * model.X'));
K = model.kernelFunction(1, 0) .^ K;
K = bsxfun(@times, model.y', K);
K = bsxfun(@times, model.alphas', K);
p = sum(K, 2);
else
% Other Non-linear kernel
for i = 1:m
prediction = 0;
for j = 1:size(model.X, 1)
prediction = prediction + ...
model.alphas(j) * model.y(j) * ...
model.kernelFunction(X(i,:)', model.X(j,:)');
end
p(i) = prediction + model.b;
end
end
% Convert predictions into 0 / 1
pred(p >= 0) = 1;
pred(p < 0) = 0;
end

192
machine-learning-ex6/ex6/svmTrain.m Normal file
View File

@@ -0,0 +1,192 @@
function [model] = svmTrain(X, Y, C, kernelFunction, ...
tol, max_passes)
%SVMTRAIN Trains an SVM classifier using a simplified version of the SMO
%algorithm.
% [model] = SVMTRAIN(X, Y, C, kernelFunction, tol, max_passes) trains an
% SVM classifier and returns trained model. X is the matrix of training
% examples. Each row is a training example, and the jth column holds the
% jth feature. Y is a column matrix containing 1 for positive examples
% and 0 for negative examples. C is the standard SVM regularization
% parameter. tol is a tolerance value used for determining equality of
% floating point numbers. max_passes controls the number of iterations
% over the dataset (without changes to alpha) before the algorithm quits.
%
% Note: This is a simplified version of the SMO algorithm for training
% SVMs. In practice, if you want to train an SVM classifier, we
% recommend using an optimized package such as:
%
% LIBSVM (http://www.csie.ntu.edu.tw/~cjlin/libsvm/)
% SVMLight (http://svmlight.joachims.org/)
%
%
if ~exist('tol', 'var') || isempty(tol)
tol = 1e-3;
end
if ~exist('max_passes', 'var') || isempty(max_passes)
max_passes = 5;
end
% Data parameters
m = size(X, 1);
n = size(X, 2);
% Map 0 to -1
Y(Y==0) = -1;
% Variables
alphas = zeros(m, 1);
b = 0;
E = zeros(m, 1);
passes = 0;
eta = 0;
L = 0;
H = 0;
% Pre-compute the Kernel Matrix since our dataset is small
% (in practice, optimized SVM packages that handle large datasets
% gracefully will _not_ do this)
%
% We have implemented optimized vectorized version of the Kernels here so
% that the svm training will run faster.
if strcmp(func2str(kernelFunction), 'linearKernel')
% Vectorized computation for the Linear Kernel
% This is equivalent to computing the kernel on every pair of examples
K = X*X';
elseif strfind(func2str(kernelFunction), 'gaussianKernel')
% Vectorized RBF Kernel
% This is equivalent to computing the kernel on every pair of examples
X2 = sum(X.^2, 2);
K = bsxfun(@plus, X2, bsxfun(@plus, X2', - 2 * (X * X')));
K = kernelFunction(1, 0) .^ K;
else
% Pre-compute the Kernel Matrix
% The following can be slow due to the lack of vectorization
K = zeros(m);
for i = 1:m
for j = i:m
K(i,j) = kernelFunction(X(i,:)', X(j,:)');
K(j,i) = K(i,j); %the matrix is symmetric
end
end
end
% Train
fprintf('\nTraining ...');
dots = 12;
while passes < max_passes,
num_changed_alphas = 0;
for i = 1:m,
% Calculate Ei = f(x(i)) - y(i) using (2).
% E(i) = b + sum (X(i, :) * (repmat(alphas.*Y,1,n).*X)') - Y(i);
E(i) = b + sum (alphas.*Y.*K(:,i)) - Y(i);
if ((Y(i)*E(i) < -tol && alphas(i) < C) || (Y(i)*E(i) > tol && alphas(i) > 0)),
% In practice, there are many heuristics one can use to select
% the i and j. In this simplified code, we select them randomly.
j = ceil(m * rand());
while j == i, % Make sure i \neq j
j = ceil(m * rand());
end
% Calculate Ej = f(x(j)) - y(j) using (2).
E(j) = b + sum (alphas.*Y.*K(:,j)) - Y(j);
% Save old alphas
alpha_i_old = alphas(i);
alpha_j_old = alphas(j);
% Compute L and H by (10) or (11).
if (Y(i) == Y(j)),
L = max(0, alphas(j) + alphas(i) - C);
H = min(C, alphas(j) + alphas(i));
else
L = max(0, alphas(j) - alphas(i));
H = min(C, C + alphas(j) - alphas(i));
end
if (L == H),
% continue to next i.
continue;
end
% Compute eta by (14).
eta = 2 * K(i,j) - K(i,i) - K(j,j);
if (eta >= 0),
% continue to next i.
continue;
end
% Compute and clip new value for alpha j using (12) and (15).
alphas(j) = alphas(j) - (Y(j) * (E(i) - E(j))) / eta;
% Clip
alphas(j) = min (H, alphas(j));
alphas(j) = max (L, alphas(j));
% Check if change in alpha is significant
if (abs(alphas(j) - alpha_j_old) < tol),
% continue to next i.
% replace anyway
alphas(j) = alpha_j_old;
continue;
end
% Determine value for alpha i using (16).
alphas(i) = alphas(i) + Y(i)*Y(j)*(alpha_j_old - alphas(j));
% Compute b1 and b2 using (17) and (18) respectively.
b1 = b - E(i) ...
- Y(i) * (alphas(i) - alpha_i_old) * K(i,j)' ...
- Y(j) * (alphas(j) - alpha_j_old) * K(i,j)';
b2 = b - E(j) ...
- Y(i) * (alphas(i) - alpha_i_old) * K(i,j)' ...
- Y(j) * (alphas(j) - alpha_j_old) * K(j,j)';
% Compute b by (19).
if (0 < alphas(i) && alphas(i) < C),
b = b1;
elseif (0 < alphas(j) && alphas(j) < C),
b = b2;
else
b = (b1+b2)/2;
end
num_changed_alphas = num_changed_alphas + 1;
end
end
if (num_changed_alphas == 0),
passes = passes + 1;
else
passes = 0;
end
fprintf('.');
dots = dots + 1;
if dots > 78
dots = 0;
fprintf('\n');
end
if exist('OCTAVE_VERSION')
fflush(stdout);
end
end
fprintf(' Done! \n\n');
% Save the model
idx = alphas > 0;
model.X= X(idx,:);
model.y= Y(idx);
model.kernelFunction = kernelFunction;
model.b= b;
model.alphas= alphas(idx);
model.w = ((alphas.*Y)'*X)';
end

24
machine-learning-ex6/ex6/visualizeBoundary.m Normal file
View File

@@ -0,0 +1,24 @@
function visualizeBoundary(X, y, model, varargin)
%VISUALIZEBOUNDARY plots a non-linear decision boundary learned by the SVM
% VISUALIZEBOUNDARYLINEAR(X, y, model) plots a non-linear decision
% boundary learned by the SVM and overlays the data on it
% Plot the training data on top of the boundary
plotData(X, y)
% Make classification predictions over a grid of values
x1plot = linspace(min(X(:,1)), max(X(:,1)), 100)';
x2plot = linspace(min(X(:,2)), max(X(:,2)), 100)';
[X1, X2] = meshgrid(x1plot, x2plot);
vals = zeros(size(X1));
for i = 1:size(X1, 2)
this_X = [X1(:, i), X2(:, i)];
vals(:, i) = svmPredict(model, this_X);
end
% Plot the SVM boundary
hold on
contour(X1, X2, vals, [0 0], 'Color', 'b');
hold off;
end

16
machine-learning-ex6/ex6/visualizeBoundaryLinear.m Normal file
View File

@@ -0,0 +1,16 @@
function visualizeBoundaryLinear(X, y, model)
%VISUALIZEBOUNDARYLINEAR plots a linear decision boundary learned by the
%SVM
% VISUALIZEBOUNDARYLINEAR(X, y, model) plots a linear decision boundary
% learned by the SVM and overlays the data on it
w = model.w;
b = model.b;
xp = linspace(min(X(:,1)), max(X(:,1)), 100);
yp = - (w(1)*xp + b)/w(2);
plotData(X, y);
hold on;
plot(xp, yp, '-b');
hold off
end

1899
machine-learning-ex6/ex6/vocab.txt Normal file
View File

File diff suppressed because it is too large Load Diff