C-BRIC Researchers at Yale Win ISLPED Best Paper Award
Congratulations to C-BRIC researchers Abhiroop Bhattacharjee, Youngeun Kim, Abhishek Moitra, and Priya Panda of Yale University.
Bhattacharjee, Kim, Moitra, and Panda won the Best Paper Award for the ACM/IEEE International Symposium on Low Power Electronics and Design (ISLPED). ISLPED is "the premier forum for presentation of innovative research in all aspects of low power electronics and design, ranging from process technologies and analog/digital circuits, simulation and synthesis tools, system-level design and optimization, to system software and applications." ISLPED is a hybrid event taking place on August 1-2, 2022 in Boston, MA and virtually.
The research team at Yale earned the award for their "Examining the Robustness of Spiking Neural Networks on Non-ideal Memristive Crossbars" publication. In this work, they examine SNNs trained via learning algorithms such as, surrogate gradient and ANN-SNN conversion. Their work showed that repetitive crossbar computations across multiple timesteps induce error accumulation, resulting in a huge performance drop during SNN inference and that SNNs trained with a smaller number of time-steps achieve better accuracy when deployed on memristive crossbars.