Sparse-Coding-Based Autoencoder and Its Application for Cancer Survivability Prediction

Algorithm 1

Proposed sparse representation-based Autoencoder algorithm.

	Input: raw training examples and , the number of maximal iterations , the stopping threshold , and regularization parameters of ();
	Initialization:
	randomly assign values to , and ;
	calculate the output from the encoder , where is the activate function;
	fortodo
(1)	Estimateandwith fixedand(using equation (12));
(2)	Estimatewith fixedandusing the ODL method;
(3)	Estimatewith fixed,and(using equation (15));
	if the predefined termination condition (the threshold ) then
	Stop training;
	end
	end
	Output: Return the optimal solution , , , and .