Bridging Neuroscience and AI: A Comprehensive Investigation of Brain-Inspired Computing Models

Artificial Intelligence, Sun Yat-Sen University, 510275 Guangzhou, China

^* Corresponding author: huim27@mail2.sysu.edu.cn

Abstract

Artificial Intelligence (AI) has reached new heights, supported by advancements in hardware and algorithm theory. Areas like robotics and autonomous driving have made significant strides, but brain-inspired computing remains a distinctive field. Although there were early hopes of AI closely connecting with brain science, this integration has been minimal. Neuroscience has mostly inspired some early algorithms, while most neural networks only adopted the idea of neuron connections without fully replicating real neural signals. However, brain-inspired algorithms, such as Spiking Neural Networks (SNNs), have shown promising results, often outperforming traditional algorithms in specific tasks and offering lower power consumption. These advancements could inspire new AI models or improve existing ones. This review explores the development of successful brain-inspired algorithms, starting with the structure and function of neurons, including cerebellar structures. It then discussed spiking neural networks, their principles, and recent research, as well as cerebellar-inspired models. Finally, the article summarizes methods for building these models and their applications in fields like robotics.