以深度学习为代表的机器学习在人工智能领域取得了重大进展，但仍存在训练时间长、算法鲁棒性差和泛化能力弱等缺陷。相比之下，大脑则能够快速、健壮地感知、学习与决策，并能举一反三。现有机器学习算法模仿了脑皮层网络的层级结构，但未体现神经元脉冲放电和突触可塑性等生物神经计算特点，这很大程度造成了其与脑智能的性能差异。本项目针对现有机器学习与人脑智能存在差异这一关键问题，综合运用计算神经科学与多模态实验数据等理论与技术，研究：1 功能脑网络的建模与控制问题；2 基于功能脑网络和生物神经计算的类脑学习算法；3 搭建类脑机器智能的实现与应用平台。项目旨在建立功能脑网络和生物神经计算联合衍生的新机器学习算法，通过类脑智能赋予机器模拟人类认知的能力，推进脑科学与类脑人工智能的融合。

Machine learning like deep learning has made significant progress in the field of artificial intelligence, but there are still shortcomings such as long training time, poor algorithm robustness and weak generalization ability. In contrast, the brain is able to perceive, learn, and make decisions quickly and robustly, and can draw inferences from simple examples. The existing machine learning algorithms mimic the hierarchical structure of the cortical network, but does not reflect the characteristics of biological nerve calculation such as pulse discharge and synaptic plasticity with neurons, which cause the performance difference from brain intelligence. This project focuses on the key problem of the difference between existing machine learning and human brain intelligence, will combine computational neuroscience and multi-modal experimental data to study: 1 Modeling and controlling of functional brain networks; 2 Designing brain-inspired learning algorithms based on functional brain networks and biological neural computing; 3 Building platforms for brain-inspired machine intelligence. Through the research of this project, a new type of machine learning algorithms derived from the combination of functional brain network and biological neural computing will be established. The brain-inspired intelligence will be further applied on robots to simulate the process of human cognition, promoting the fusion of brain science and brain-inspired artificial intelligence.