Chinese Scientists Unveil Brain-Inspired Neural Circuit Evolution
Chinese scientists have recently unveiled an innovative strategy for brain-inspired neural circuit evolution, presenting exciting possibilities for developing spiking neural networks that closely resemble biological systems in terms of efficiency and plausibility. The study, led by Zeng Yi from the Institute of Automation under the Chinese Academy of Sciences (CAS), proposes a groundbreaking approach to advancing general brain-inspired cognitive intelligence.
The human brain consists of various neural circuits that contribute to different cognitive functions, working together harmoniously to enhance human perception, learning, and decision-making. By exploring the dynamic characteristics of these biological neural circuits from a computational perspective, researchers aim to leverage this knowledge to enhance artificial intelligence systems.
Through the application of evolved neural circuits, the team of scientists successfully constructs spiking neural networks designed for image classification and reinforcement learning tasks. Leveraging the brain-inspired Neural circuit Evolution strategy (NeuEvo) together with diverse neural circuit types, the evolved spiking neural network demonstrates significantly enhanced capabilities in perception and reinforcement learning tasks, according to the study.
Incorporating on-policy and off-policy deep reinforcement learning algorithms, NeuEvo achieves performance levels comparable to artificial neural networks, offering promising results for future applications. These findings were recently published in Proceedings of the National Academy of Sciences (PNAS), a well-respected scientific journal in the United States.
This breakthrough research opens up new avenues for developing artificial intelligence systems that imitate the brain’s intricate neural circuits more closely. By leveraging the knowledge gained from studying the dynamic characteristics of biological neural circuits, Chinese scientists are harnessing this computational perspective to enhance the capabilities of AI systems. The evolved spiking neural networks display remarkable improvements in perception and reinforcement learning tasks, paving the way for more efficient and biologically plausible AI models.
The implications of this study stretch beyond artificial intelligence systems. By better understanding the adaptive synergy among different neural circuits in the brain, scientists can gain insights into human cognition. This knowledge can be applied in various fields, including psychology, neuroscience, and even healthcare, potentially leading to advancements in treatments for cognitive disorders and brain injuries.
The Chinese scientists’ research sheds light on the immense potential of brain-inspired neural circuit evolution and its applications in the field of artificial intelligence. As the study demonstrates the remarkable improvements achieved by incorporating evolved neural circuits into spiking neural networks, it paves the way for more biologically plausible and efficient AI systems. With further advancements and collaborations within the scientific community, we can expect to witness groundbreaking developments in the field of brain-inspired cognitive intelligence.
