Recently, Dr. Li Guoqi, a researcher at the Laboratory of Brain Atlas and Brain-Inspired Intelligence, Institute of Automation, Chinese Academy of Sciences, was selected by DeepTech as one of the “2022 China Intelligent Computing Science and Technology Innovation Figures,” receiving high recognition for his academic contributions and technological innovations in the field of brain-inspired computing.

According to reports, the DeepTech “2022 China Intelligent Computing Science and Technology Innovation Figure” award focuses on computing for intelligence, intelligence-driven computing, and intelligent/data-driven scientific discovery. It aims to identify and highlight the individuals behind technological innovations, practical applications, and valuable real-world impact, creating a benchmark for technology and talent in the intelligent computing industry, showcasing the latest academic research results and technological breakthroughs, and promoting progress in intelligent computing science and technology.

Dr. Li Guoqi has devoted himself to research on brain-inspired computing theory, architecture, and software–hardware co-design. In brain-inspired modeling, he was a core member in the development of Tianjixin, the world’s first heterogeneously integrated brain-inspired chip, where he addressed theoretical problems such as network training compression and compilation deployment of mapping algorithms on brain-inspired hardware. As a key team member, he also co-proposed a hierarchical brain-inspired system architecture based on “brain-inspired completeness,” decoupling software and hardware design and extending the classical “Turing completeness” definition to brain-inspired systems.

In brain-inspired architecture, he systematically reviewed the key technologies and development trends of AI chip architectures combined with model compression, mapping, and hardware acceleration. On this basis, he optimized a heterogeneous, multi-core, parallel software–hardware co-designed brain-inspired architecture. In brain-inspired algorithms, he proposed efficient training algorithms for spiking neural networks (SNNs) targeting brain-inspired chips, solving the problem of spike degradation (gradient vanishing) during SNN training and providing algorithmic support for the widespread application of low-power brain-inspired chips. This series of work also addressed the technical challenges of training large-scale, ultra-deep SNNs and deploying them in heterogeneous integration with artificial neural networks (ANNs).

He is the principal investigator of an NSFC Key Project titled “Brain-Inspired Spiking Neural Network Online Learning Model and Architecture Integrating Attention Mechanisms,” which focuses on brain-inspired computing theory and architecture. He also leads the NSFC Regional Innovation Joint Key Project “Heterogeneous Integrated Brain-Inspired Computing Chip and System,” aiming to advance the industrialization of brain-inspired chips and systems.