README.md

In EdwinGraham/EdNet: Train deep neural networks for classification and regression

EdNet

Train deep neural networks with fully connected layers for classification and regression (Vignette.)

Installation

devtools::install_github("EdwinGraham/EdNet")

Features

• Train models for classification tasks
• Binary
• Multi-class
• Regression supported for following distribtuion
• Normal
• Poisson
• Gamma
• Tweedie
• Supports weighted regression
• Supports training with an offset model
• L1 and L2 regularisation
• Drop-out
• Mini-batch learning
• Train with the following optimisation algorithms
• Gradient descent
• Momentum
• RMSProp
• Adam
• Initialise with random weights or continue learning from existing checkpoint model.

EdwinGraham/EdNet documentation built on May 6, 2019, 12:22 p.m.