In response to the challenges of balancing accuracy and interpretability in existing stock price prediction models,a hiera-rchical neural fuzzy network model for stock price prediction was proposed.An adaptive neural fuzzy network unit incorporating an attention mechanism(ANFIS-A)was introduced,and a hierarchical adaptive neural fuzzy network was constructed using these units.A feature subset selection algorithm was proposed by combining the binary grey wolf optimization algorithm(BG-WO).A recursive rule elimination algorithm was introduced to further reduce the number of rules and enhance their interpreta-bility.Experimental results demonstrate that this model exhibits high accuracy and interpretability in stock price prediction.
关键词
灰狼优化算法/层次自适应模糊神经网络/注意力机制/股票价格预测/可解释性/金融时间序列/规则消除
Key words
grey wolf optimization algorithm/hierarchical adaptive neuro-fuzzy inference system(H-AHFIS)/attention mecha-nism/stock price prediction/interpretability/financial time series/rule elimination
维普
万方数据