NEAL: Neuromorphic Embodied Agents that Learn

by Chris Huyck

A few years ago we made some video game agents from simulated neurons. The most advanced was the Cell Assembly roBot version 3 ( CABot3).

The first goal of NEAL is to port CABot3 from our fatiguing leaky integrate and fire neural model (with 10 ms time steps) to PyNN and then onto SpiNNaker and HICANN.

The second goal is to improve the learning of Cell Assemblies in general, and of visual categories in particular, and of plans.