A novelty psychological cognition behaviour model based on reinforcement learning

Abstract

The goal of the paper is to effectively illustrate psychological cognition behaviours using the reinforcement learning. Combined with the trust behaviour of human society, an reinforcement learning model based on human trust habits is put forward: 1) self-adaptive overall knowability decision-making method based on the historical evidence window is constructed, which not only has overcome the subjective judgment method for the determination of weights commonly used in existing models, but also can solve the knowability forecast problem when the direct evidence is insufficient; 2) The concept of reinforcement learning weighted averaging (hereinafter referred to as RLWA for short) operator is introduced, and the direct trust forecast model based on the RLWA operator is established, which can be sued to solve the problem of insufficient dynamic adaptability of the traditional forecast model. The experimental results show that, compared with the existing models, the proposed model has more robust dynamic adaptability and also significant improvement in the forecast accuracy of the model.