Stock market price forecasting based on DPSO-LSTM hyperparameter tuning
Long Short-Term Memory(LSTM)is suitable for processing and predicting important events with very long intervals and delays in time series.It is a challenging task to manually find an efficient network configuration due to its complex network structure,uncertain hyperparameters,and time-consuming network training.In this paper,Distributed Particle Swarm Optimization(DPSO)is used to effectively solve the hy-perparameter tuning problem of LSTM,and research the selection of the optimal number of hidden ele-ments,activation function and learning rate in LSTM to find a high-performance LSTM.Based on the histor-ical transaction data of CSI300 for price prediction,the experiment results show that the method is effective,which provides new ideas and methods for hyperparameter tuning and stock market price prediction.
万方数据
维普
