Theme 1: Stochastic Information Processing Systems

Objective

The Stochastic Information Processing Systems (SIPS) theme seeks to establish statistical foundations for information processing to achieve energy efficiency and robustness in emerging nanoscale beyond-CMOS functional fabrics, which are physically stochastic. It will do so by matching the fabrics’ inherent statistical attributes to those of emerging machine learning frameworks for statistical inference (which are functionally stochastic). The goals of the SIPS theme are to achieve orders-of-magnitude improvements in energy-efficiency and robustness for compute-intensive, sensing, and communication-intensive tasks implemented in beyond-CMOS technologies.

Faculty

Malik, Olshausen, Rutenbar, Shanbhag, Singer (Theme Lead), Varshney, Madhow, Wornell, Verma, Mitra