Data Scientist, CTO Office and Intel AI Lab
Xin received this PhD training in the field of systems and computational neuroscience at the University of Southern California. His PhD thesis answered a few important outstanding questions in subcortical visual information processing, specifically, principles by which neural circuits process sensory information encoded in spike trains. Since 2010, Xin joined Terry Sejnowski’s lab as a postdoctoral fellow and later a research collaborator at the Salk Institute. His work during this period addressed a number of outstanding questions on the function and pathology of cortical circuit dynamics. His research was recognized by the Life Sciences Foundation that awarded him a Pfizer Fellowship, in addition to a Young Investigator Award from the Brain and Behavior Research Foundation afterwards. From 2013 to 2014, Xin spent a two-year stint in Corporate R&D at Qualcomm as a senior systems engineer, conducting research projects in developing a neuromorphic processor. In collaboration with hardware engineers, Xin invented a number of event-based sensing and learning algorithms of potential neuromorphic applications, which resulted in numerous US and European patents. Since 2017, Xin joined Intel-Nervana, now Artificial Intelligence Products Group (AIPG) at Intel as a data scientist, conducting cutting-edge research and development of technologies for specialized hardware for modern deep learning. Recently, Xin transitioned to a research role in the Office of the CTO, focusing on the next generation of computing technology beyond conventional hardware architecture and beyond deep learning.
Network pruning has emerged as a powerful technique for reducing the size of deep neural networks. Pruning uncovers high-performance subnetworks…
Deep neural networks are commonly developed and trained in 32-bit floating point format. Significant gains in performance and energy efficiency…
Sorry, there was an error in your submission.