Brain-inspired computing, which aims to develop artificial intelligence (AI) via learning from neural systems, is receiving increasing attention in both academia and industry. Deep Learning, as the main driving force of the current wave of AI, mimics the feedforward, the hierarchical, and the parallel processing characteristics of visual information processing. However, a real biological visual system has much more complicated structures and holds much richer computational powers. In this talk, I will review particularly those dynamical aspects of visual information processing that are missed in the current framework of Deep Learning, and they are inspirational for us to develop artificial general intelligence (AGI).
Biography
Prof. Si Wu is a PI from the State Key Lab of Cognitive Neuroscience & Learning, Beijing Normal University. His research interests are Computational Neuroscience and Machine Learning. His lab is using mathematical models and large-scale computer simulations to elucidate brain functions, and meanwhile, developing brain-style information processing techniques. The projects currently going on in his group include: continuous attractor neural networks for information representation, computational roles of short-term synaptic plasticity, neural mechanisms for multi-sensory integration, and brain-inspired algorithms for visual information processing. Their research results have been published in top journals or prestigious conferences in the field, including Neuron, Nature Neuroscience, PNAS, J. Neurosci., NIPS et al.