ziemowit-s / neuronpp_gym

Spiking Neural Gym agent created on top of NEURON++ and OpenAI Gym libraries

Spiking Neural Gym agent created on top of NEURON++ and OpenAI Gym libraries

Prerequisites

• Python 3.6

• Install requirements.txt

• Install NEURON from the instruction: https://github.com/ziemowit-s/neuron_get_started

• Install OpenAI Gym Atari environments

  pip install gym[atari]

Run

• pong_gym_run.py

Example Usage

Example of Atari PONG game (left) and visual receptive fields inputs (right) to pixels' changes

Known issues:

• After a long run it may terminate with:

terminate called after throwing an instance of 'std::bad_alloc'
  what():  std::bad_alloc

Process finished with exit code 134 (interrupted by signal 6: SIGABRT)