Neat-Ex

v1.3.0

Neat-Ex v1.3.0 Ann.Simulation

A module for simulating ANNs.

Summary

Functions

change(sim)

Returns the ammount of data change in the latest step of the simulation

eval(sim, inputs, max \\ 100, thresh \\ 1.0e-5)

Uses a single frame of inputs repeatedly until it has evaluated max times, or until the summation of data changes is <= to thresh

exportAnn(sim)

Applies the weightMap back onto the ANN and returns it. Note, this is only of use if the stored weightMap has been modified manually or by an external source such as the Backprop or NeuralTrainer modules

new(ann, sigmoid \\ fn x -> :math.tanh(2 * x) end)

Creates a new ANN Simulation given an ANN, and a sigmoid function

step(sim, inputs)

Steps the simulation forward one frame, provided with the current inputs for this frame. inputs should be a map of node_ids and their values

Functions

change(sim)

Returns the ammount of data change in the latest step of the simulation.

eval(sim, inputs, max \\ 100, thresh \\ 1.0e-5)

Uses a single frame of inputs repeatedly until it has evaluated max times, or until the summation of data changes is <= to thresh.

exportAnn(sim)

Applies the weightMap back onto the ANN and returns it. Note, this is only of use if the stored weightMap has been modified manually or by an external source such as the Backprop or NeuralTrainer modules.

new(ann, sigmoid \\ fn x -> :math.tanh(2 * x) end)

Creates a new ANN Simulation given an ANN, and a sigmoid function.

step(sim, inputs)

Steps the simulation forward one frame, provided with the current inputs for this frame. inputs should be a map of node_ids and their values.