Skip to main content
Module

x/neat_ai/mod.ts>NeatOptions

DenoJS Neural network project written in typescript.
stSoftwareAU/NEAT-AI
Go to Latest
interface NeatOptions
implements TrainOptions
import { type NeatOptions } from "https://deno.land/x/neat_ai@v0.54.0/mod.ts";

Properties

optional
costName: string
optional
creativeThinkingConnectionCount: number

How many new links to create during the creative thinking phase

optional
creatureStore: string

The directory to store the creatures (optional)

optional
dataSetPartitionBreak: number

number of records per dataset file. default: 2000

optional
debug: boolean

debug (much slower)

optional
experimentStore: string

The directory to store the experiments (optional)

optional
creatures: CreatureInternal[] | CreatureExport[]

List of creatures to start with

optional
feedbackLoop: boolean

Feedback loop ( previous result feeds back into next interaction https://www.mathworks.com/help/deeplearning/ug/design-time-series-narx-feedback-neural-networks.html;jsessionid=2d7fa2c64f0bd39c86dec46870cd

optional
focusList: number[]

The list of observations to focus one

optional
focusRate: number

Focus rate

optional
costOfGrowth: number

Cost of growth

optional
elitism: number
optional
timeoutMinutes: number

Once the number of minutes are reached exit the loop.

optional
trainPerGen: number

the number of training per generation. default: 1

optional
maxConns: number

The maximum number of connections

optional
maximumNumberOfNodes: number

The maximum number of nodes

optional
mutationAmount: number

Number of changes per Gene

optional
mutationRate: number

Probability of changing a gene

optional
populationSize: number

The target population size. 50 by default

optional
threads: number

the number of workers

optional
selection: SelectionInterface
optional
mutation: MutationInterface[]
optional
iterations: number
optional
log: number
optional
verbose: boolean

verbose logging default: false