NeuPy v0.6.0

NeuPy is a Python library for Artificial Neural Networks. NeuPy supports many different types of Neural Networks from a simple perceptron to deep learning models.

Installation

$ pip install neupy

User Guide

• Install NeuPy
• Check the tutorials
• Learn more about NeuPy in the documentation
• Explore lots of different neural network algorithms.
• Read articles and learn more about Neural Networks.

Links

• Tutorials
• Documentation
• Articles
• Cheat sheet
• Open Issues

Articles

• Hyperparameter optimization for Neural Networks
• Visualize Backpropagation Algorithms
• MNIST classification
• Predict prices for houses in the area of Boston
• Password recovery
• Discrete Hopfield Network

Jupyter Notebooks

• Hyperparameter optimization for Neural Networks
• Playing with MLP visualizations
• Visualizing CNN based on Pre-trained VGG19
• Looking inside of the VGG19 using SOFM

Examples

Convolutional Neural Networks

Image classification

• MNIST CNN
• CIFAR10 CNN
• Pre-trained AlexNet CNN
• Pre-trained VGG16 CNN
• Pre-trained VGG19 CNN
• Pre-trained SqueezeNet
• GoogleNet
• Inception v3
• ResNet 50

Reinforcement Learning

• Neural Network plays CartPole game
• Value Iteration Network (VIN)

Recurrent Neural Networks

• Reber Grammar Classification, sequence input and fixed-size output with GRU
• Generate Shakespeare text, sequence input and sequence output with LSTM

Autoencoders

• MNIST, Denoising Autoencoder
• MNIST, Convolutional Autoencoder
• MNIST, Stacked Convolutional Autoencoders
• MNIST, Variational Autoencoder

Boltzmann Machine

• Feature Learning from the MNIST Images, Restricted Boltzmann Machine (RBM)
• Gibbs sampling using face images, Restricted Boltzmann Machine (RBM)

MLP Neural Networks

Regression

• Boston House Price prediction, Hessian algorithm

Image classification

• MNIST, Multilayer perceptron
• Rectangle images, Multilayer perceptron

Visualizations

• Learning Algorithms Visualization, Gradient Descent, Momentum, RPROP and Conjugate Gradient

Classification

• IMDB review classification using CBOW and RPROP MLP
• MLP with categorical and numerical features

Competitive Neural Networks

SOFM Clustering

• Simple SOFM example
• Clustering iris dataset using SOFM

SOFM Data topology learning

• Learning half-circle topology with SOFM

Explore SOFM features

• Compare feature grid types for SOFM
• Compare weight initialization methods for SOFM

SOFM Visualizations

• Visualize digit images in 2D space with SOFM
• Embedding 30-dimensional dataset into 2D and building heatmap visualization for SOFM

LVQ

• Reduce number of training samples in iris dataset with LVQ3

Neural Networks with Radial Basis Functions

• Classify iris dataset, Probabilistic Neural Network (PNN)
• Regression using Diabetes dataset, Generilized Neural Nerwork (GRNN)
• Music-Speech audio classification, Probabilistic Neural Network (PNN)

Memory based Neural Networks

• Sinus function approximation, CMAC
• Visualize Discrete Hopfield Neural Network energy function
• Password recovery, Discrete Hopfield Neural Network

Dependencies

• Python 2.7, 3.4, 3.5, 3.6
• Theano == 0.9.0
• NumPy >= 1.9.0
• SciPy >= 0.19.0
• Matplotlib >= 1.4.0
• graphviz == 0.5.1