Lifelong learning using spiking neural networks

Abstract: Deep learning has made tremendous progress in the last ten years but it's high computational and memory requirements impose challenges for its expanding application base, particularly for edge devices. Further, leveraging the information present in the differential data seen by edge devices requires development of new efficient distributed learning algorithms. There has been some progress in lowering memory requirements of deep neural networks (for instance, use of half-precision) but there has been minimal effort in exploring alternative efficient computational paradigms. Inspired by the brain, Spiking Neural Networks (SNN) provide an energy-efficient alternative to conventional rate-based neural networks. However, SNN architectures that employ the traditional feedforward and feedback pass do not fully exploit the asynchronous event-based processing paradigm of SNNs. In my talk, I will present recently developed methods within my group that provide energy and memory efficient alternatives for lifelong distributed learning and generative modeling. These methods rely on biological plasticity mechanism of predictive coding to develop networks that can simultaneously perform multiple tasks like classification and generation.

Bio: Shirin Dora is a lecturer in the Department of Computer Science at Loughborough University. Prior to joining Loughborough University, he briefly worked as a lecturer in the Intelligent Systems Research Centre at the Ulster University in Northern Ireland. He received his PhD from Nanyang Technological University, Singapore in 2017. During his PhD, he focused on developing energy-efficient learning algorithms for spiking neural networks using inspirations from the brain. After his PhD, he joined the Cognitive and Systems Neuroscience Group at the University of Amsterdam as a postdoctoral researcher. In Amsterdam, he delved deeper into the plasticity mechanisms in the brain specifically focusing on predictive coding for perception and multisensory integration. His current research interests include efficient techniques for lifelong learning, few-shot learning and distributed AI.

Organiser: S Fatima