Speaker
Prof. Bo Lei, Researcher

Chair
Prof. Shan Yu

Time
10:00, Thursday, September 4th, 2025
(Note: The English part on the poster mistakenly writes “Tuesday, April 15th, 2025,” the correct date is Thursday, September 4th, 2025.)

Venue
No. 3 Meeting Room (3rd Floor), Intelligent Building,
Institute of Automation, Chinese Academy of Sciences

Abstract

As one of the core disciplines for analyzing the principles of biological and medical phenomena, neuroscience has received tremendous investment and rapid development. With this, neuroscience research is also ushering in an era of data explosion. However, the construction of general models of human intelligence at the data and application levels, as well as in theoretical neuroscience, still lacks effective technical approaches. The core difficulty lies in the large individual differences present in neuroscience research. Collecting data across modalities and tasks, together with the large-scale data-driven training of artificial intelligence models, poses unique challenges.

To address this, we have proposed a series of solutions: by building large-scale multimodal datasets, we can achieve “cross-subject” and “cross-modality” integration of different neural activity data and develop predictive models with generalized cognitive ability. This allows us to test the “Scaling Law” of neuroscience with cross-subject and cross-task data.

In this talk, we will introduce the preliminary construction of a neuroscience foundation model, Brainu. This model integrates fMRI, EEG, MEG, optical imaging, physiological signals, and other sources of multimodal data. It enables single-model processing of multiple modalities and integrates neural representations from different scenarios. Brainu is expected to become a powerful tool for data analysis in fundamental neuroscience research and can be applied to auxiliary diagnosis of brain diseases and enhancement of brain-computer interface applications.

Biography

Prof. Bo Lei is a researcher at Beijing Academy of Artificial Intelligence, head of the AI + Neuroscience research direction. He is responsible for fundamental AI models for neuroscience and next-generation large-scale biological data–driven artificial neural network research.

As first author or corresponding author, he has published in Nature, Neuron, PNAS, Nature Communications, Science Advances, Cell Reports, ICML, and other top journals and conferences. One of his works was selected in “Top Ten Scientific Advances in China, 2020.”

Lecture No. 012-20250904-03

Organizer: Laboratory of Brain Atlas and Brain-inspired Intelligence