Annex v0.2.1 API Reference

Modules

Annex is a library for composing and running deep artificial

The Cost module defines the types, callbacks, and helper functions for calculating a the loss and gradient of the lossd, for gradient descent, of a network.

MeanSquaredError is the module that encapsulates the calculation of the mean squared error equation.

Annex.Data defines the callbacks and helpers for data structures used by Annex.

A 2D Matrix Annex.Data that uses the Tensor library as the underlying data structure.

The Annex.Data.List is the most basic Annex.Data.

List2D is a 2 dimensional list of lists of floats.

A module for running checks and getting helpful messages.

Defaults handles the access to compile-time configuration values for default values of Annex.

The Annex.Layer is the module that defines types, callbacks, and helper for Layers.

The Activation layer is the Annex.Layer that is responsible for applying an activation function to the data during the feedforward and supplying the gradient function (derivative) of the activation function to the Backprops during backpropagation.

Backprop is a module that contains helper functions for manipulating the Keyword list that is passed up through a Sequence during the backprop phase.

rows are the number outputs and columns are the number of inputs.

Given a frequency the dropout layer randomly drops an input at the frequency.

The Sequence layer is the container and orchestrator of other layers and is used to define a list of Layers that compose a deep neural network.

The Annex.LayerConfig is the intermediate structure used to intialize an Annex.Layer.

The Learner module defines the types, callbacks, and helper functions for a Learner.

The Optimizer Behaviour and context for calling optimizer implementations.

The optimizer for mini-batch (or non-batching) stochastic gradient descent.

A simple perceptron Learner capable of making good predictions given a linearly separable dataset and labels.

The Shape module encapsulates helper functions for use in determining the shapes and validity of shapes between Layers and Layers; Data and Data; and Data and Layers.

This module holds functions that are not necessarily specific to one of Annex's other modules.

Exceptions