sentimentron/nnet

A small collection of neural network algorithms in Go

NNET - Neural Networks in Go

NNET is a small collection of neural network algorithms written in the pure Go language.

Packages

• rbm - Binary-Binary Restricted Boltzmann Machines (RBMs)
• gbrbm - Gaussian-Binary RBMs
• mlp - Multi-Layer Perceptron (Feed Forward Neural Networks)
• mlp3 - Three-Layer Perceptron
• dbn - Deep Belief Nets (in develop stage)

Install

go get github.com/r9y9/nnet

Examples

Binary-Binary Restricted Bolztmann Machines on MNIST

cd examples/rbm
go run rbm_mnist.go -h # for help
go run rbm_mnist.go -epoch=5 -hidden_units=400 -learning_rate=0.1 -order=1 -output="rbm.json" -persistent -size=20

It took 32 minutes to train RBM on my macbook air at 07/28/2014.

Weight visualization

python visualize.py rbm.json

Multi-layer Perceptron

Training

cd examples/mlp3
go run mlp3_mnist.go -epoch=500000 -hidden_units=100 -learning_rate=0.1 -o="nn.json"

It took 10 minutes to train MLP on my macbook air at 07/30/2014.

Classification

go run mlp3_mnist.go -test -m=nn.json
...
Acc. 0.971000 (9710/10000)

TODO

• Use linear algebra library such as gonum/matrix or go.matrix
• GPU powered training
• Refactor (write more idiomatic codes, speedup, etc.)
• Tests for all packages
• More flexibility like pylearn2

License

MIT