ECE Spring Seminar Series
Friday, May 18, 2018
2:00 PM-3:30 PM
Architecture for Scalable Neuromorphic Systems
Abstract: Neural network is a well-studied area of artificial intelligence and has matured over the past decades. Recently, there is a revived interest in neural networks due to the following developments: (1) advances in material technology such as non-volatile memories enable implementation of neural networks in resource and power-constrained devices, (2) growing volume of data from internet and sensory nodes (big data) and growing size of neural networks (big models) for high dimensional problems limit the scalability of conventional design of training and inference in hardware, and (3) new application areas present new challenges, both in terms of neural network model development (in software) and their efficient hardware implementation (neuromorphic computing). This talk will give an overview of research challenges on neuromorphic computing, including use of non-volatile memory as synapses. Evolving computing paradigms such as in-memory computing and distributed processing will be discussed to evaluate its potential for building a large scale neuromorphic computing platform.
Bio: Dr. Anup Das (M’14) is an Assistant Professor in the Department of Electrical and Computer Engineering at Drexel University. He obtained his Ph.D. in Embedded Systems from the National University of Singapore in 2014. He was a researcher in the Neuromorphic Computing Group of IMEC, Netherlands (2015 - 2017) where he was leading a group of scientists for the development of scalable neuromorphic systems as part of the EU Human Brain Project. He was a post-doctoral fellow in the School of Electronics and Computer Science (ECS) at the University of Southampton, UK (2014 - 2015). Between 2004 and 2011, he worked at LSI Corporation and STMicroelectronics, as senior IC design engineer. His research interests include design of algorithms for neuromorphic computing, particularly using spiking neural networks, dataflow-based design of neuromorphic computing system, design of scalable computing system; hardware-software co-design and management, and thermal and power management of many-core embedded systems.