/SMARTAgent

Primary LanguagePythonMIT LicenseMIT

SMARTAgent

A closed-loop spiking neuronal network (visual/motor) model that senses dynamic visual information from the AIgame enviroment and learns to produce actions that maximize game reward through spike-timing dependent reinforcement learning.

Reference

Training a spiking neuronal network model of visual-motor cortex to play a virtual racket-ball game using reinforcement learning https://doi.org/10.1371/journal.pone.0265808