In this paper, a hybrid evolutionary intelligent system is proposed for dimensionality reduction and tuning the learnable parameters of artificial neural network (ANN) that can forecast the future (1‐day‐ahead) close price of the stock market using various technical indicators. Although the ANN possesses the ability to model highly uncertain and complex nonlinear data but the key challenge in ANN is tuning its parameters and minimizing the feature set that can be used in the input layer. The backpropagation approach used for training the ANN has a limitation to get trapped in local minima and overfitting the data. Motivated by this, we proposed a hybrid intelligent system for optimizing the initial parameters and for reducing the dimensions of the feature set. The proposed model is a combination of feature extraction technique, namely principal component analysis (PCA), particle swarm optimization (PSO), and Levenberg‐Marquardt (LM) algorithm for training the feed‐forward neural networks (FFNN). This paper also compares the forecasting efficiency of the proposed model with PSO‐FFNN, regular FFNN, two standard benchmark approaches viz. GA and DE and another hybrid model obtained by the combination of PCA and a time series econometric model viz. auto‐regressive distributed lag model. The presented technique has been tested to predict the close price of three stock indices viz. Nifty 50, Sensex, and S&P 500. Simulation results indicate that the proposed model shows superior forecasting accuracy as compared with other methods.
In this paper, a two-stage swarm intelligence based hybrid feed-forward neural network approach is designed for optimal feature selection and joint optimization of trainable parameters of neural networks in order to forecast the close price of Nifty 50, Sensex, S&P 500, DAX and SSE Composite Index for multiple-horizon (1-day ahead, 5-days-ahead and 10-days ahead) forecasting. Although the neural network can deal with complex non-linear and uncertain data but defining its architecture in terms of number of input features in the input layer, the number of neurons in the hidden layer and optimizing the weights is a challenging problem. The backpropagation algorithm is frequently used in the neural network and has a drawback to getting stuck in local minima and overfitting the data. Motivated by this, we introduce a swarm intelligence based hybrid neural network model for automatic search of features and other hlearnable neural networks' parameters. The proposed model is a combination of discrete particle swarm optimization (DPSO), particle swarm optimization (PSO) and Levenberg-Marquardt algorithm (LM) for training the feedforward neural networks. The DPSO attempts to search automatically the optimum number of features and the optimum number of neurons in the hidden layer of FFNN whereas PSO, simultaneously tune the weights and bias in different layers of FFNN. This paper also compares the forecasting efficiency of proposed model with another hybrid model obtained by integrating binary coded genetic algorithm and real coded genetic algorithm with FFNN. Simulation results indicate that the proposed model is effective for obtaining the optimized feature subset and network structure and also shows superior forecasting accuracy.
In this paper, we presented a long short-term memory (LSTM) network and adaptive particle swarm optimization (PSO)based hybrid deep learning model for forecasting the stock price of three major stock indices such as Sensex, S&P 500, and Nifty 50 for short term and long term. Although the LSTM can handle uncertain, sequential, and nonlinear data, the biggest challenge in it is optimizing its weights and bias. The back-propagation through time algorithm has a drawback to overfit the data and being stuck in local minima. Thus, we proposed PSO-based hybrid deep learning model for evolving the initial weights of LSTM and fully connected layer (FCL). Furthermore, we introduced an adaptive approach for improving the inertia coefficient of PSO using the velocity of particles. The proposed method is an aggregation of adaptive PSO and Adam optimizer for training the LSTM. The adaptive PSO attempts to evolve the initial weights in different layers of the LSTM network and FCL. This research also compares the forecasting efficacy of the proposed method to the genetic algorithm (GA)-based hybrid LSTM model, the Elman neural network (ENN), and standard LSTM. Experimental findings demonstrate that the suggested model is successful in achieving the optimum initial weights and bias of the LSTM and FC layers, as well as superior forecasting accuracy.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.