Title:Energy-based Probing for Spiking Neural Networks
Deep neural networks have made significant strides in various domains, but they often require substantial computational resources due to extensive matrix multiplication even for inference. As a result, energy consumption becomes a concern especially for mobile devices. Spiking neural networks (SNNs) have garnered increasing attention due to their potential for low-energy real-world applications. However, SNNs currently face two primary challenges: the non-differentiability of the spike firing function and the performance gap between artificial neural networks (ANNs) and SNNs. In this talk, I will discuss our efforts to address these challenges and present potential solutions.
Bin Gu is currently an assistant professor in the Department of Machine Learning at the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI). Prior to joining MBZUAI, he held the position of full professor at Nanjing University of Information Science and Technology. His previous research interests primarily revolved on large-scale optimization in machine learning and data mining. However, his current focus lies in the area of large-scale optimization for Spiking Neural Networks and its applications. He has published 90 more papers, with over 3,000 citations according to Google Scholar. He served as a program committee member or reviewer for several leading machine learning and data mining conferences and journals such as NeurIPS, ICML, KDD, AAAI, TPAMI, JMLR, and a senior program committee member of IJCAI 2019-2021, AAAI 2023.