LSTMs have been shown to learn many difficult sequential tasks effectively, including speech recognition, machine translation, trajectory prediction, and correlation analysis ( Elman, 1990 Jordan, 1990 Hochreiter and Schmidhuber, 1997 Schuster and Paliwal, 1997 Cho et al., 2014 Alahi et al., 2016 Zhou et al., 2016 Su et al., 2017 Gupta et al., 2018 Hasan et al., 2018 Li and Cao, 2018 Salman et al., 2018 Vemula et al., 2018 Xu et al., 2018 Yang et al., 2019). To alleviate this issue, the variants of RNNs with gating mechanisms, such as long short-term memory (LSTM) networks and gated recurrent units (GRU), have been proposed. However, the errors signal back-propagated through time often suffer from exponential growth or decay, a dilemma commonly referred to as exploding or vanishing gradient. In RNNs, the recurrent layers or hidden layers consist of recurrent cells, and whose states are affected by both past states and current input with feedback connections ( Yu et al., 2019). With the rapid development of artificial intelligence and machine learning, the recurrent neural network (RNN) models have been gaining interest as a statistical tool for dealing with the complexities of sequential data ( Chung et al., 2015 Keren and Schuller, 2016 Sadeghian et al., 2019 Yang et al., 2019). Many types of information (e.g., language, music, and gene) can be represented as sequential data that often contains related information separated by many time steps, and these long-term dependencies are difficult to model as we must retain information from the whole sequence with greater complexity of the model ( Trinh et al., 2018 Liu et al., 2019 Shewalkar, 2019 Yu et al., 2019 Zhao et al., 2020). By comparing the average accuracy of real datasets with long short-term memory, Bi-LSTM, gated recurrent units, and MCNN and calculating the main indexes (Accuracy, Precision, Recall, and F1-score), it can be observed that our method can improve the average accuracy and optimize the structure of the recurrent neural network and effectively solve the problems of exploding and vanishing gradients.ĭata classification is one of the most important tasks for different applications, such as text categorization, tone recognition, image classification, microarray gene expression, and protein structure prediction ( Choi et al., 2017 Johnson and Zhang, 2017 Malhotra et al., 2017 Aggarwal et al., 2018 Fang et al., 2018 Mikołajczyk and Grochowski, 2018 Kerkeni et al., 2019 Saritas and Yasar, 2019 Yildirim et al., 2019 Chandrasekar et al., 2020). It provides six pathways so as to fully and deeply explore the effect and influence of historical information on the RNNs. For each method, there are two ways for historical information addition: 1) direct addition and 2) adding weight weighting and function mapping to activation function. To include the historical information, we design two different processing methods for the SS-RNN in continuous and discontinuous ways, respectively. At the same time, for the time direction, it can improve the correlation of states at different moments. It can enhance the long-term memory ability. ![]() To solve these problems, this paper proposes a new algorithm called SS-RNN, which directly uses multiple historical information to predict the current time information. However, they have problems such as insufficient memory ability and difficulty in gradient back propagation. Recurrent neural networks are widely used in time series prediction and classification.
0 Comments
Leave a Reply. |