Neural Networks in the Wolfram Language

The Wolfram Language has state-of-the-art capabilities for the construction, training and deployment of neural network machine learning systems. Many standard layer types are available and are assembled symbolically into a network, which can then be immediately trained and deployed on available CPUs and GPUs.

Introduction

LeNet and MNIST

Layers

Net Encoders

Net Decoders

Containers

Training

Evaluation

Advanced Concepts

Training on Large Datasets

Out-of-Core Training on MNIST

Using a MongoDB Database

Example Weighting

Regularization

Classification

Classifying Vector Data

Classifying Categorical Data

Multitask Classification

Regression

Predicting from Numeric Data

Regression with Uncertainty

Computer Vision

Image Classification

MNIST Digit Classification

CIFAR-10 Object Classification

Learned Embedding

Style Transfer

Semantic Segmentation (Pixel Classification)

Semantic Segmentation on a Toy Text Dataset

Sequence Learning and NLP

Sequence Regression

Integer Addition

Sequence Classification

Sentiment Analysis

Sequence-to-Sequence Learning

Fixed-Length Integer Addition

Variable-Length Integer Addition

Integer Sorting

Optical Character Recognition (OCR) on a Toy Dataset

Question Answering

Simple RNN Trained on the bAbI QA Dataset

Memory Network Trained on the bAbI QA Dataset

Language Modeling

Character-Level Language Model

Audio Analysis

Audio Encoding

Encoder Types

Data Augmentation

Audio Classification

Audio Event Detection

Audio Embeddings

Siamese Networks with Contrastive Loss

Pre-trained Audio Feature Extractors

Transfer Learning for Audio

Unsupervised Learning

Manifold Learning

Autoencoder

MNIST Digit Autoencoder