CM1K Neuromorphic Chip Emulator

This project offers a Python emulator of General Vision's CM1K neuromorphic chip. The chip implements a Restricted Coulomb Energy / Radial Basis Function network in hardware using 1024 artificial neurons. A pattern of up to 256 bytes in length can be presented to the neurons in parallel, essentially realizing a constant-time classification relative to the network size. Furthermore, multiple CM1K chips can be modularly connected into networks of a million neurons while continuing to incur essentially no increase in computation time (a million neurons can classify a pattern in very nearly the same time as a single chip of 1K neurons).

The emulator includes routines for ingesting four popular public datasets (MNIST, AT&T Faces, Iris, Mushroom) into the network and subsequently running statistically rigorous experiments against those datasets to characterize the network's performance.

Please check out this slide presentation that describes neuromorphic computing in general, with an emphasis on the CM1K, and then walks through a series of experiments I ran using the emulator to evaluate the CM1K's performance.

The CM1K emulator is available on GitHub at https://github.com/kebwi/CM1K_emulator.