Inspirations from Neuroscience for Brain-Inspired Computing
Prof. Sen Song
Department of Biomedical Engineering
School of Medicine
Date & Time
CB-A, G/F, Chow Yei Ching Building, HKU
Currently, we stand at an interesting time in history. Scaling in element size and computing power in VLSI electronics driven by Moore’s law is hitting the fundamental barrier in physics. Recent rapid progress from deep learning is also plateauing and the weakness of this approach is now becoming apparent. Both endeavors could have much to learn from the brain, which still has far superior general intelligence and power efficiency than its electronic counterparts. Recent technological advances have brought neuroscience into a big-data era. However, a general theory of intelligence is yet to emerge. I will introduce some of the key findings and challenges of brain emerging from recent neuroscience research on the three levels of single synapse and neuron, local circuits, and large scale brain-wide circuits. I will also try to highlight a few key principles that might offer inspirations for brain-inspired computing.
Prof. Sen Song is the Assistant Director of Tsinghua Laboratory of Brain and Intelligence and a Principal Investigator at Department of Biomedical Engineering, School of Medicine, Tsinghua University. He is also a member of the Center for Brain-inspired Computing Research Center, Beijing Innovation Center for Future Chips, Institute of Artificial Intelligence, and McGovern Institute for Brain Research.
He received PhD degree in computational neuroscience from Brandeis University in 2002, and completed postdoctoral research at Cold Spring Harbor Laboratory and Massachusetts Institute of Technology. In 2010, he joined the Department of Biomedical Engineering at Tsinghua University.His main research interest is in interdisciplinary research in computational neuroscience, brain-inspired computing and artificial intelligence. He also has considerable experience in research on neural circuits, bioinformatics, and genomics. His representative work includes the theoretical study on the spike-timing-dependent plasticity (STDP) published in Nature Neuroscience and Neuron, and the study on the motif analysis of local brain circuits published in PLoS Biology. In recent years, his lab has also conducted a series of work on neural circuits underlying emotion and motivation, which are published in Cell Reports, Journal of Neuroscience and others. In the field of artificial intelligence, he is interested in applying deep learning to the analysis of brain imaging data and healthcare and education. His recent new interests also include: brain-scale cognitive models, modeling and analysis of complex temporal and spatial data such as EEG with applications in social interaction and positive psychology research, and neuro-aesthetics.