Evolutionary Reasoning
Peijun Ye, Fei–Yue Wang · 2023
Evolutionary reasoning aims to simulate human's possible and heterogenous deliberation trajectories. This chapter addresses the reasoning issue. It discusses knowledge representation, which is the fundamental “bricks” of the cognitive “building.” The chapter explains the process of reasoning. The reasoning follows the bottom–up emergent principle of complex biological systems. It is built upon evolutionary computing. The chapter then focuses on the second key point of evolutionary reasoning – the fitness function. Since the arbitrary setting of fitness function may bring individual bias in traditional evolutionary computing, it introduces inverse reinforcement learning to adaptively and automatically determine the fitness functions. Knowledge representation is a comprehensive and fundamental issue. The evolutionary reasoning is composed of three steps: initial solution generation, mutation, and selection.