摘 要：
针对下一代的神经网络设计，本研究以模拟人脑工作方式为目标，探索了生物神经元的能量优化与树突计算功能模拟两个重要方面。首先，我们提出了一种独特的生物神经元能量优化模型。这一模型依据神经元内部的生物化学反应调控神经元的电流流动，以此达到降低能耗，提高能源使用效率的目的。初步研究结果显示，这种模型的能源利用效率比传统神经网络提高了约20％。然后，我们对树突的计算功能进行了模拟。与神经元的输入、输出相比，树突计算的建模更为复杂，但同时也具有潜力实现多种复杂的计算任务。在此基础上，我们设计了第一代树突计算模型，并通过实验验证了其计算能力。整体而言，对生物神经元的模拟研究，不仅对神经科学的发展有所贡献，也为人工智能的进一步发展指明了一条新的理论研究道路。
关键词：生物神经元; 能量优化; 树突计算
Abstract:
For the design of the next-generation neural networks, this study aims to simulate the working mechanisms of the human brain and explores two crucial aspects: energy optimization of biological neurons and simulation of dendritic computation functions. Firstly, we propose a unique energy optimization model for biological neurons. This model regulates the current flow within neurons based on their internal biochemical reactions, thereby achieving the goals of reducing energy consumption and improving energy utilization efficiency. Preliminary research results indicate that this model enhances energy utilization efficiency by approximately 20% compared to traditional neural networks. Subsequently, we simulate the computational functions of dendrites. Compared to the input and output of neurons, modeling dendritic computation is more complex but also holds the potential to accomplish a variety of complex computational tasks. On this basis, we design the first-generation dendritic computation model and experimentally verify its computational capabilities. Overall, the simulation research on biological neurons not only contributes to the advancement of neuroscience but also charts a new theoretical research path for the further development of artificial intelligence.
Keywords: Biological neurons; Energy optimization; Dendritic computation
--
