DMelt:AI/4 Recurrent NN and LSTM

From jWork.ORG
Jump to: navigation, search
Limitted access. Reguest membership or login to this link first if you are already a member
Contents

Recurrent Neural Networks (RNN) and LSTM

Recurrent Neural Networks (RNN) and Long Short-Term Memory Networks (LSTM) are included using recunn/ Java library under the MIT license. Recurrent networks and their popular type, LSTM, are artificial neural network designed to recognize patterns in sequences of data, such as text, sequences, handwriting, the spoken words, or numerical times series.

Here is a demo that reads sentences from Paul Graham's essays, encoding Paul Graham's knowledge into the weights of the Recurrent Networks. The long-term goal of the project then is to generate startup wisdom at will. Feel free to train on whatever data you wish, and to experiment with the parameters. If you want more impressive models you have to increase the sizes of hidden layers, and maybe slightly the letter vectors.

The above example is based on JavaScript library http://cs.stanford.edu/people/karpathy/recurrentjs/

More information on this topic is in DMelt books