Anticipatory Learning Classifiers in Python - PyALCS

Anticipatory Learning Classifier Systems (ALCS)

ALCS are is an extension to basic LCS compromising the notation of anticipations. Doing that the systems predominantly are able to anticipate perceptual consequences of actions independent of a reinforcement predictions.

ALCS are able to form complete anticipatory representation (build environment model) which allows faster and intelligent adaptation of behaviour or problem classification.

Guide

Contents:

• Installation
  • Creating environment with Conda (example)
  • Dependencies
  • Launching example integrations
  • Interactive notebooks
• ACS2
  • Original code
  • Maze statistics
• rACS
  • Real-value representation
  • Changes
• Jupyter notebooks
  • Description of ACS2
  • ACS2 in Maze
  • ACS2 in Frozen Lake