PI: Hal Finkel, LCF

Objective:  Neuromorphic Computing Hardware is fast becoming a reality, but how to best train and make use of these artificial neural networks in hardware is still an open question. This project has a dual focus.

Testbed descriptions:
  1. Test potential use cases for neuromorphic hardware in high-performance computing, bioinformatics, simulation data analysis, intelligent data acquisition devices (e.g. detectors in high-energy physics experiments), and more. The goal is to develop a quantitative understanding of the requirements in these areas.
2. Conduct research on the training techniques best used for these devices, especially focusing on using evolutionary algorithms (e.g. genetic algorithms) as replacements to, or enhancements of, existing techniques.