time series prediction with deep neural network

1 view (last 30 days)
i saw many examples on classification with deep neural network . i am working on time series prediction , as deep learning has revolutionized everything i want to forecast my data with deep neural networks . i tried it with 2 hidden layers of stacked auto encoder by fine tunning . but it is not working .i m getting very bad predictions here is the code that i have used
i changed the log sigmoid to tansig and the last layer /output layer function from softmax to purelin
i also changed the training fuction from trainscg to trainlam as compared to classification problem . because in first autoencoder with scg it was not learning anything the gradients were NaN and the final output was just a single horizontal line on 0 axis.
i trieed to increase number of hidden neurons but it given out of memory errors kindly tell me how to do it with batch learning. how to break it into mini size batches.
anyone knows hoe to do deep learning with RBM or stacked autoencoder for predictions or forecasting problems
thankyou

Answers (0)

Categories

Find more on Sequence and Numeric Feature Data Workflows in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!