www.gusucode.com > nnet 案例源码 matlab代码程序 > nnet/digittrain_dataset.m
function [inputs,targets] = digittrain_dataset % digittrain_dataset Synthetic handwritten digit dataset for training % [inputs, targets] = digittrain_dataset will load a dataset of synthetic % handwritten digits, where inputs will be a 1-by-5000 cell array, with % each cell containing a 28-by-28 matrix representing a synthetic image % of a handwritten digit, and targets will be a 10-by-5000 matrix % containing the labels for the images in 1-of-K format. % % Example: % Train a sparse autoencoder on images of handwritten digits and % visualize the learned features. % % x = digittrain_dataset; % hiddenSize = 25; % autoenc = trainAutoencoder(x, hiddenSize, ... % 'L2WeightRegularization', 0.004, ... % 'SparsityRegularization', 4, ... % 'SparsityProportion', 0.2); % plotWeights(autoenc); % % See also digittest_dataset, digitsmall_dataset % Copyright 2014-2015 The MathWorks, Inc. load digittrain_dataset; inputs = xTrainImages; targets = tTrain; end