Chinese scientists have newly proposed a strategy for brain-inspired neural circuit evolution, shedding light on developing spiking neural networks with better biological plausibility and efficiency, according to the Chinese Academy of Sciences (CAS).

This study proposes an idea for the research of general brain-inspired cognitive intelligence, said Zeng Yi, a researcher with the Institute of Automation under the CAS and leader of the study team.

Different neural circuits in the brain are involved in the development of different cognitive functions, and the adaptive synergy of the circuits promotes human perception, learning, and decision-making. It is of value to explore the dynamic characteristics of biological neural circuits from a computational perspective and apply them to enhance the capabilities of artificial intelligence systems.

By applying the evolved neural circuits, the researchers construct spiking neural networks for image classification and reinforcement learning tasks. Using the brain-inspired Neural circuit Evolution strategy (NeuEvo) with rich neural circuit types, the evolved spiking neural network greatly enhances capability on perception and reinforcement learning tasks, according to the study.

Combined with on-policy and off-policy deep reinforcement learning algorithms, NeuEvo achieves comparable performance with artificial neural networks, as shown in the study.

The study was recently published in the U.S. journal Proceedings of the National Academy of Sciences.