PyRNN

LSTM-RNN implementation. For more details about usage check examples folder.

Usage

1. Create own class with superclass ProblemObject and inherit its methods
2. Create RNN using

rnn = RNN(input_dimension, hidden_dimension, output_dimension, alpha = 0.1) 3. Learn RNN with dataset. Learining is supervized. X is list of problemObjects

rnn.learn(X)

4. Use learned RNN. To predict values of problemObject with unknown result.

rnn.predict(problem)

TODO

• Visualisation of convergence
• Bad input checking
• Add more examples
• Parallelisation on Spark

Acknowledgement

I want to thank Andrew Trask for great tutorial which I used during implementation. RNN tutorial