Learning scripts

kmmbvnr · Mar 23, 2012 · 542ae45 · 542ae45
1 parent 98f5d7d
commit 542ae45
Show file tree

Hide file tree

Showing 42 changed files with 6,127 additions and 3 deletions.
diff --git a/CryDetector/src/cc/wthr/crydetector/CryDetector.java b/CryDetector/src/cc/wthr/crydetector/CryDetector.java
@@ -72,7 +72,7 @@ public void link(int sessionId) {
 				throw new RuntimeException(e);
 			}
 		}
-
+				
 		mEqualizer = new Equalizer(0, sessionId);
         mEqualizer.setEnabled(true);
 
@@ -81,7 +81,7 @@ public void link(int sessionId) {
 		mVisualizer.setDataCaptureListener(this, Visualizer.getMaxCaptureRate(), false, true);
 		mVisualizer.setEnabled(true);
 	}
-	
+
 	public void unlink() {
 		if(mVisualizer != null) {			
 			mVisualizer.setEnabled(false);

diff --git a/CryDetector/src/cc/wthr/crydetector/MainActivity.java b/CryDetector/src/cc/wthr/crydetector/MainActivity.java
@@ -106,7 +106,7 @@ public void onClick(View view) {
 			try { 
 				mPlayer = MediaPlayer.create(this, soundId);
 				mCryDetector.link(mPlayer.getAudioSessionId());
-				mPlayer.setOnCompletionListener(this);
+				mPlayer.setOnCompletionListener(this);				
 				mPlayer.start();
 			} catch(Throwable e) {
 				Log.d("MainActivity", "Link visualizer error", e);	

diff --git a/CryTeacher/logs/cry1.txt b/CryTeacher/logs/cry1.txt
diff --git a/CryTeacher/logs/cry10.txt b/CryTeacher/logs/cry10.txt
diff --git a/CryTeacher/logs/cry11.txt b/CryTeacher/logs/cry11.txt
diff --git a/CryTeacher/logs/cry12.txt b/CryTeacher/logs/cry12.txt
diff --git a/CryTeacher/logs/cry15.txt b/CryTeacher/logs/cry15.txt
diff --git a/CryTeacher/logs/cry2.txt b/CryTeacher/logs/cry2.txt
diff --git a/CryTeacher/logs/cry3.txt b/CryTeacher/logs/cry3.txt
diff --git a/CryTeacher/logs/cry5.txt b/CryTeacher/logs/cry5.txt
diff --git a/CryTeacher/logs/cry6.txt b/CryTeacher/logs/cry6.txt
diff --git a/CryTeacher/logs/cry7.txt b/CryTeacher/logs/cry7.txt
diff --git a/CryTeacher/logs/cry9.txt b/CryTeacher/logs/cry9.txt
diff --git a/CryTeacher/logs/nocry1.txt b/CryTeacher/logs/nocry1.txt
diff --git a/CryTeacher/logs/nocry10.txt b/CryTeacher/logs/nocry10.txt
diff --git a/CryTeacher/logs/nocry11.txt b/CryTeacher/logs/nocry11.txt
diff --git a/CryTeacher/logs/nocry12.txt b/CryTeacher/logs/nocry12.txt
diff --git a/CryTeacher/logs/nocry13.txt b/CryTeacher/logs/nocry13.txt
diff --git a/CryTeacher/logs/nocry14.txt b/CryTeacher/logs/nocry14.txt
diff --git a/CryTeacher/logs/nocry15.txt b/CryTeacher/logs/nocry15.txt
diff --git a/CryTeacher/logs/nocry17.txt b/CryTeacher/logs/nocry17.txt
diff --git a/CryTeacher/logs/nocry3.txt b/CryTeacher/logs/nocry3.txt
diff --git a/CryTeacher/logs/nocry4.txt b/CryTeacher/logs/nocry4.txt
diff --git a/CryTeacher/logs/nocry5.txt b/CryTeacher/logs/nocry5.txt
diff --git a/CryTeacher/logs/nocry6.txt b/CryTeacher/logs/nocry6.txt
diff --git a/CryTeacher/logs/nocry7.txt b/CryTeacher/logs/nocry7.txt
diff --git a/CryTeacher/logs/nocry8.txt b/CryTeacher/logs/nocry8.txt
diff --git a/CryTeacher/logs/nocry9.txt b/CryTeacher/logs/nocry9.txt
diff --git a/CryTeacher/matlab/costFunction.m b/CryTeacher/matlab/costFunction.m
@@ -0,0 +1,33 @@
+function [J, grad] = costFunction(theta, X, y)
+%COSTFUNCTION Compute cost and gradient for logistic regression
+%   J = COSTFUNCTION(theta, X, y) computes the cost of using theta as the
+%   parameter for logistic regression and the gradient of the cost
+%   w.r.t. to the parameters.
+
+% Initialize some useful values
+m = length(y); % number of training examples
+
+% You need to return the following variables correctly 
+J = 0;
+grad = zeros(size(theta));
+
+% ====================== YOUR CODE HERE ======================
+% Instructions: Compute the cost of a particular choice of theta.
+%               You should set J to the cost.
+%               Compute the partial derivatives and set grad to the partial
+%               derivatives of the cost w.r.t. each parameter in theta
+%
+% Note: grad should have the same dimensions as theta
+%
+
+
+J = 1/m * ( -y' * log(sigmoid(X*theta)) - (1-y)' * log(1 - sigmoid(X*theta)));
+
+grad = X' * (sigmoid(X*theta)-y) ./ m;
+
+
+
+
+% =============================================================
+
+end
diff --git a/CryTeacher/matlab/costFunctionReg.m b/CryTeacher/matlab/costFunctionReg.m
@@ -0,0 +1,30 @@
+function [J, grad] = costFunctionReg(theta, X, y, lambda)
+%COSTFUNCTIONREG Compute cost and gradient for logistic regression with regularization
+%   J = COSTFUNCTIONREG(theta, X, y, lambda) computes the cost of using
+%   theta as the parameter for regularized logistic regression and the
+%   gradient of the cost w.r.t. to the parameters. 
+
+% Initialize some useful values
+m = length(y); % number of training examples
+
+% You need to return the following variables correctly 
+J = 0;
+grad = zeros(size(theta));
+
+% ====================== YOUR CODE HERE ======================
+% Instructions: Compute the cost of a particular choice of theta.
+%               You should set J to the cost.
+%               Compute the partial derivatives and set grad to the partial
+%               derivatives of the cost w.r.t. each parameter in theta
+
+% do not regularize on the Theta0
+theta_1 = theta;
+theta_1(1) = 0;
+
+J = 1/m * ( -y' * log(sigmoid(X*theta)) - (1-y)' * log(1 - sigmoid(X*theta))) + lambda/(2*m) * sum (theta_1 .* theta_1);
+
+grad = (X' * (sigmoid(X*theta)-y) + lambda .* theta_1)./ m;
+
+% =============================================================
+
+end
diff --git a/CryTeacher/matlab/data.txt b/CryTeacher/matlab/data.txt
diff --git a/CryTeacher/matlab/featureNormalize.m b/CryTeacher/matlab/featureNormalize.m
@@ -0,0 +1,17 @@
+function [X_norm, mu, sigma] = featureNormalize(X)
+%FEATURENORMALIZE Normalizes the features in X 
+%   FEATURENORMALIZE(X) returns a normalized version of X where
+%   the mean value of each feature is 0 and the standard deviation
+%   is 1. This is often a good preprocessing step to do when
+%   working with learning algorithms.
+
+mu = mean(X);
+X_norm = bsxfun(@minus, X, mu);
+
+sigma = std(X_norm);
+X_norm = bsxfun(@rdivide, X_norm, sigma);
+
+
+% ============================================================
+
+end
diff --git a/CryTeacher/matlab/predict.m b/CryTeacher/matlab/predict.m
@@ -0,0 +1,25 @@
+function p = predict(theta, X)
+%PREDICT Predict whether the label is 0 or 1 using learned logistic 
+%regression parameters theta
+%   p = PREDICT(theta, X) computes the predictions for X using a 
+%   threshold at 0.5 (i.e., if sigmoid(theta'*x) >= 0.5, predict 1)
+
+m = size(X, 1); % Number of training examples
+
+% You need to return the following variables correctly
+p = zeros(m, 1);
+
+% ====================== YOUR CODE HERE ======================
+% Instructions: Complete the following code to make predictions using
+%               your learned logistic regression parameters. 
+%               You should set p to a vector of 0's and 1's
+%
+
+p = sigmoid(X*theta) >= 0.5;
+
+
+
+% =========================================================================
+
+
+end
diff --git a/CryTeacher/matlab/sigmoid.m b/CryTeacher/matlab/sigmoid.m
@@ -0,0 +1,18 @@
+function g = sigmoid(z)
+%SIGMOID Compute sigmoid functoon
+%   J = SIGMOID(z) computes the sigmoid of z.
+
+% You need to return the following variables correctly 
+g = zeros(size(z));
+
+% ====================== YOUR CODE HERE ======================
+% Instructions: Compute the sigmoid of each value of z (z can be a matrix,
+%               vector or scalar).
+
+
+g = 1 ./ (1 + exp(-1 .* z));
+
+
+% =============================================================
+
+end
diff --git a/CryTeacher/matlab/train.m b/CryTeacher/matlab/train.m
@@ -0,0 +1,98 @@
+% Cry Detector Training
+
+data = load('data.txt');
+
+% Split data to train, test and crossvalidation sets
+total_count = size(data,1);
+data = data(randperm(total_count), :);
+
+train_count = int32(total_count*0.6);
+test_count = int32(total_count*0.2);
+cv_count = total_count - train_count - test_count;
+
+
+Xtrain = data(1:train_count, 1:128);
+ytrain = data(1:train_count, 129);
+
+Xtest = data(train_count+1:train_count+test_count, 1:128);
+ytest = data(train_count+1:train_count+test_count, 129);
+rtest = data(train_count+1:train_count+test_count, 130);
+
+Xval = data(train_count+test_count+1:total_count, 1:128);
+yval = data(train_count+test_count+1:total_count, 129);
+
+X = Xtrain;
+y = ytrain;
+
+% Xval = featureNormalize(Xval);
+% yval = featureNormalize(yval);
+
+#lambda_vec = [0, 0.1, 0.5, 1, 10, 50, 100 1000 5000 10000 30000 50000 60000 100000]';
+lambda_vec = [1000 5000 10000 30000]';
+#error_train = zeros(length(lambda_vec), 1);
+#error_val = zeros(length(lambda_vec), 1);
+
+[m, n] = size(X);
+X = [ones(m, 1) X];                % Add intercept term to X
+Xval = [ones(cv_count, 1) Xval];                % Add intercept term to Xval
+Xtest = [ones(test_count, 1) Xtest];
+
+% Compute and display initial cost and gradient
+initial_theta = zeros(n + 1, 1); % Initialize fitting parameters
+[cost, grad] = costFunctionReg(initial_theta, X, y, 0);
+fprintf('Cost at initial theta (zeros): %f\n', cost);
+
+options = optimset('GradObj', 'on', 'MaxIter', 400);
+[theta, cost] = fminunc(@(t)(costFunctionReg(t, X, y, 0)), initial_theta, options);
+
+% error_train = zeros(size(1:50:m,2), 1);
+% error_val   = zeros(size(1:50:m,2), 1);
+% for k=1:50:m
+%    fprintf('%d\n', k);
+%    Xtrain = X(1:k,:);
+%    ytrain = y(1:k);
+%    [theta, cost] = fminunc(@(t)(costFunctionReg(t, Xtrain, ytrain, 0)), initial_theta, options);
+%    error_train(int32(k/50)+1) = 100 * (1 - mean(predict(theta, Xtrain) == ytrain));
+%    error_val(int32(k/50)+1) = 100 * (1 - mean(predict(theta, Xval) == yval));
+%end
+
+% plot(1:50:m, error_train, 1:50:m, error_val);
+% title('Learning curve for linear regression')
+% legend('Train', 'Cross Validation')
+% xlabel('Number of training examples')
+% ylabel('Error')
+
+% Selecting right lambda
+% options = optimset('GradObj', 'on', 'MaxIter', 10000);
+% for k = 1:length(lambda_vec)
+%   lambda = lambda_vec(k);
+%   [theta, cost] = fminunc(@(t)(costFunctionReg(t, X, y, lambda)), initial_theta, options);
+%  error_train(k) = 100 * (1 - mean(predict(theta, X) == y));
+%  error_val(k) = 100 * (1 - mean(predict(theta, Xval) == yval));
+%  error_train(k)
+%  error_val(k)
+%  fprintf('%f Cost at theta found by fminunc: %f\n', lambda, cost);
+%end
+
+% plot(lambda_vec, error_train, lambda_vec, error_val);
+% title('Learning curve for linear regression')
+% legend('Train', 'Cross Validation')
+% xlabel('lambda')
+% ylabel('Error')
+
+
+% Compute accuracy on our training set
+p = predict(theta, Xtest);
+
+fprintf('Train Accuracy: %f \n', mean(double(p == ytest)) * 100);
+fprintf('Total size %d \n', length(rtest == 1));
+fprintf('Total right predicted %d \n', sum(predict(theta, Xtest) == ytest));
+fprintf('Right predicted cries: %f \n', sum((predict(theta, Xtest) == ytest)(ytest==1)));
+fprintf('Right predicted nocries: %f \n', sum((predict(theta, Xtest) == ytest)(ytest==0)));
+fprintf('Not predicted cries: %f \n', sum((predict(theta, Xtest) != ytest)(ytest==1)));
+fprintf('Wrong predicted cries: %f \n', sum((predict(theta, Xtest) != ytest)(ytest==0)));
+fprintf('Total Cry %d\n', sum(ytest));
+fprintf('Total NoCry %d\n', sum(ytest==0));
+pause
+
+theta