进化大规模脉冲神经网络的发育方法

A Developmental Method for Evolving Large-Scale Spiking Neural Networks

通过自然进化得到的脑包含几十亿的神经元和几万亿的神经连接,并表现出复杂的智能行为.受生物脑进化与发育的启发,研究者给出了进化神经网络的发育编码方法,特点是通过基因重用可在较小的基因空间中进行大规模神经网络的快速搜索.以人工基因组模型为框架描述基因调控网络,用基因表达的动态特性表示细胞命运特化的发育过程,提出了一种进化大规模脉冲神经网络的发育方法.该方法的特点在于可以快速有效地发育生成脉冲神经元、神经连接和突触可塑性.相应的食物采集进化实验突现了以神经驱动的自主智能体的智能行为,并验证了该方法对大规模脉冲神经网络的进化能力.

The brains contain billions of neurons and trillions of connections through natural evo- lution, and appear rather complex intelligent behavior. Researchers in developmental encoding, which is a branch of evolutionary neural networks motivated by the evolution and development of biological brains, often point out that genetic reuse allows searching the large-scale neural net- works through a lower dimensional genotypic space. Using the artificial genome model as a framework for describing genetic regulatory networks, the dynamics of gene expression can be treated as a model for cell fate specification. We propose a developmental method for evolving large-scale spiking neural networks. The advantage of this method is that it can facilitate fast and efficient development of spiking neurons, neural connections, and synaptic plasticities. The cor- responding evolutionary experiment shows that the intelligent behavior emergences for the neu- rally-driven autonomous agents in a food gathering task. Additionally, it also shows that due to the efficiency of the proposed method, large-scale spiking neural networks can be easily managed thereby making it suitable for long durational evolutionary experiments.