Javascript Machine Learning Library.
Viktor is available as an NPM module
npm install --save viktor
Documentation for Viktor can be found at here
Examples of Viktor in use: MNIST
Neural Networks are implemented using the stones library's Matrix object, and currently supports configuration of nodes and the learning rate. The current network defaults to the Softmax model and a sigmoid activation function as it was built to be optimized for MNIST.
As there are many ways to build neural networks, the next version of viktor will include more models and activation functions. Then, performance optimizations using GPU computation will be next, followed by Network wrappers to allow optimization of learning rates during training
Stanford Deep Learning Cheat Sheet Stanford Supervised Learning Cheat Sheet Stanford Unsupervised Learning Cheat Sheet Stanford Tips and Tricks