search

danijar / layered

Clean implementation of feed forward neural networks
MIT License
240 stars 33 forks source link

Implement batch backpropagation as matrix times matrix multiplication #6

Closed danijar closed 8 years ago

danijar commented 8 years ago

Do a single matrix multiplication for the training batch rather than looping over examples and performing vector multiplications. This also solves allows the Gradient interface to accept either single examples or batches.

danijar commented 8 years ago

Not needed anymore since activation derivatives get the above derivative as a parameter now.