neon: A highly configurable deep learning framework

Release:0.8.1 Date:May 06, 2015

Contents

• Quick start
  • Getting started
  • Machine learning OPerations (MOP) Layer
  • Upcoming libraries
  • License
• Installation
  • Overview
  • Required Dependencies
  • Optional Dependencies
  • Configuration Setup
  • Installing MPI on an Ubuntu cluster (for distributed models)
  • Virtualenv
  • Upgrading
  • Uninstalling
• Using neon
  • Running
  • Key neon Command-line Arguments
  • Experiment File Format
  • Parallelization
  • Training Models and Learning Parameters
  • Generating Predictions
  • Reporting Performance
• Distributed Implementations using MPI
  • Available Models
  • Examples
• Hyperparameter optimization
  • Installing our Spearmint Fork
  • YAML file changes
  • Experiment Initialization
  • Running
• Contributing to the Framework
  • neon Contribution Process
  • Architecture
  • Extending the Framework
• ML OPerational Layer (MOP) API
  • neon.backends.backend.Tensor
  • neon.backends.backend.Backend
  • Basic Data Structure
  • Arithmetic Operation Support
  • Logical Operation Support
  • Summarization Operation Support
  • Initialization and Setup
  • Higher Level Operation Support
• MOP API Changes
  • v0.9.0
  • v0.8.0
  • v0.7.0
• Backends
  • Current Implementations
  • Adding a new Backend
• Models
  • Available Models
  • Adding a new type of Model
• Layers
  • Available Layers
  • Adding a new type of Layer
• Transforms
  • Available Transforms
• Metrics
  • Reporting Metric Values
  • Available Metrics
• Learning Rules
  • Available Learning Rules
  • Adding a new type of Learning Rule
• Datasets
  • Available Datasets
  • Basic YAML Parameters
  • Adding a new type of Dataset
  • Working with Imageset
• Experiments
  • Current Implementations
  • Adding a new type of Experiment
  • Generating Predictions
• Utility Functions
  • Python 2 and 3 Compatibility
  • Persistence of objects and data
  • Batched Data Writing
• Frequently Asked Questions
  • Installation
  • Running
  • Contributing
• API Reference
  • Architecture
  • API Functions
  • CPU
  • Cudanet GPU
  • Nervana GPU
  • Nervana Hardware
  • MLP
  • Autoencoder
  • Balance Network
  • RBM
  • DBN
  • Recurrent Neural Network
  • Cost Layer
  • Activation Layer
  • Data Layer
  • Weight Layer
  • Fully Connected Layer
  • Convolutional Layer
  • Pooling Layers
  • DropOut Layer
  • Composite Layers
  • Normalized Layers
  • Boltzmann Layers
  • Recurrent Layers
  • Gradient Descent
  • Value Initialization
  • Misclassification
  • ROC
  • Loss
  • Squared Error
  • Activation Functions
  • Cost Functions
  • MNIST
  • CIFAR10
  • Iris
  • Sparsenet
  • Mobydick
  • Synthetic

Indices and tables

• Index
• Module Index
• Search Page