1st working version of NYTRO crossval. Needs further testing.

config_set.m

@@ -1,7 +1,56 @@
-function [ output_args ] = config_set( input_args )
-%CONFIG_SET Summary of this function goes here
-%   Detailed explanation goes here
+function [ config ] = config_set( varargin )
+%CONFIG_SET Constructs the default  configuration stucture to be used by
+%nytro_train
 
+    % Set default configuration fields
+    config = struct();
 
+    % data
+    config.data.shuffle = 1;
+
+    % crossValidation
+    config.crossValidation.storeTrainingError = 0;
+    config.crossValidation.validationPart = 0.2;
+    config.crossValidation.recompute = 0;
+    config.crossValidation.errorFunction = @rmse;
+    config.crossValidation.codingFunction = [];
+    config.crossValidation.stoppingRule = @windowLinearFitting;
+    config.crossValidation.windowSize = 10;
+    config.crossValidation.threshold = 0;
+
+    % filter
+    config.filter.fixedIterations  = [];
+    config.filter.maxIterations  = 500;
+    config.filter.gamma  = [];
+
+    % kernel
+    config.kernel.kernelFunction  = @gaussianKernel;
+    config.kernel.kernelParameters = 1;
+    config.kernel.m = 100;
+
+    % Parse function inputs
+    if ~isempty(varargin)
+
+        % Assign parsed parameters to object properties
+        fields = varargin(1:2:end);
+        for idx = 1:numel(fields)
+
+            currField = fields{idx};
+            % Parse current field
+            k = strfind(currField , '.');
+            k = [0 ; k ; (numel(currField)+1)];
+            tokens = cell(1,(numel(k) - 1));
+            for i = 1 : (numel(k) - 1);
+                tokens{i} = currField( (k(i)+1) : (k(i+1)-1) );
+            end
+
+            cmdStr = 'config';
+            for i = 1 : (numel(tokens) - 1)
+                cmdStr = strcat(cmdStr , '.(''' , tokens{i} , ''')');
+            end
+            cmdStr = strcat(cmdStr , '.(''' , tokens{end} , ''') = varargin{2*(idx-1) + 2};');
+            eval(cmdStr);
+        end
+    end
 end
 

examples/simpleUsage.m

@@ -8,4 +8,4 @@
 [ training_output ] = nytro_train( Xtr , Ytr );
 
 % Perform predictions on the test set and evaluate results
-[ prediction_output ] = nytro_test( Xte , Yte , training_output);
+[ prediction_output ] = nytro_test( Xtr , Xte , Yte , training_output);

nytro_test.m

@@ -11,7 +11,7 @@
 
     % Perform prediction
     tic
-    output.YtePred = KnmTe * training_output.config.best.alpha;
+    output.YtePred = KnmTe * training_output.best.alpha;
     if ~isempty(training_output.config.crossValidation.codingFunction)
         output.YtePred = codingFunction(output.YtePred);
     end

nytro_train.m

@@ -105,17 +105,17 @@
         output.best.t = Inf;
 
         % Error buffers
-        if config.crossValidation.storeValidationError == 1
-            output.errorPath.validation = zeros(1,config.filter.maxIterations) * NaN;
-        else
-            output.errorPath.training = [];
-        end
+        output.errorPath.validation = zeros(1,config.filter.maxIterations) * NaN;
         if config.crossValidation.storeTrainingError == 1
             output.errorPath.training = zeros(1,config.filter.maxIterations) * NaN;
         else
             output.errorPath.training = [];
         end
 
+        % Init time structures
+        output.time.crossValidationTrain = 0;
+        output.time.crossValidationEval = 0;
+
         % Subdivide training set in training1 and validation
 
         ntr1 = floor( ntr * ( 1 - config.crossValidation.validationPart ));
@@ -146,7 +146,7 @@
 
         % Compute kernel
         tic
-        Knm = kernelFunction(Xtr, Xtr1(nysIdx,:), config.kernel.kernelParameters);
+        Knm = config.kernel.kernelFunction(X, Xtr1(nysIdx,:), config.kernel.kernelParameters);
         Kmm = Knm(trainIdx(nysIdx),:);
         R = chol( ( Kmm + Kmm') / 2 + 1e-10 * eye(config.kernel.m));  % Compute upper Cholesky factor of Kmm
         output.time.kernelComputation = toc;
@@ -206,7 +206,7 @@
                 stop  = config.crossValidation.stoppingRule(...
                             output.errorPath.validation(1:iter) , ...
                             config.crossValidation.windowSize , ...
-                            config.crossValidation.thres);
+                            config.crossValidation.threshold);
 
                 if stop == 1
                     break
@@ -216,7 +216,7 @@
             output.time.crossValidationTotal = output.time.crossValidationEval + output.time.crossValidationTrain;
         end    
 
-        if config.crossValidation.retraining == 1
+        if config.crossValidation.recompute == 1
 
             %%% Retrain on whole dataset
 
@@ -253,7 +253,7 @@
 
         % Compute kernels
         tic
-        Knm = kernelFunction(X, X(nysIdx,:), config.kernel.kernelParameters);
+        Knm = config.kernel.kernelFunction(X, X(nysIdx,:), config.kernel.kernelParameters);
         Kmm = Knm(nysIdx,:);
         R = chol( ( Kmm + Kmm') / 2 + 1e-10 * eye(config.kernel.m));  % Compute upper Cholesky factor of Kmm    
         output.time.kernelComputation = toc;