Hybrid Long Short-Term Memory prediction model improved by particle swarm optimization with sine and cosine factors

The Long Short-Term Memory network of deep learning neural is widely used to predict stock price in financial field. In order optimize the accuracy prediction by LSTM network, this paper firstly uses principal component analysis method extract various influencing indexes stock. Then, use Circle mapping select initial value more evenly, sine and cosine factors improve particle swarm optimization algorithm, so as find optimal parameters model effectively. Finally, results IPSO algorithm are substituted into for regression with components. Through empirical comparative test, show that improved proposed has better effect, not prone local problems, based on higher accuracy.