Applications of Recurrent Neural Networks in Environmental Factor Forecasting: A Review

Abstract

Analysis and forecasting of sequential data, key problems in various domains of engineering and science, have attracted the attention of many researchers from different communities. When predicting the future probability of events using time series, recurrent neural networks (RNNs) are an effective tool that have the learning ability of feedforward neural networks and expand their expression ability using dynamic equations. Moreover, RNNs are able to model several computational structures. Researchers have developed various RNNs with different architectures and topologies. To summarize the work of RNNs in forecasting and provide guidelines for modeling and novel applications in future studies, this review focuses on applications of RNNs for time series forecasting in environmental factor forecasting. We present the structure, processing flow, and advantages of RNNs and analyze the applications of various RNNs in time series forecasting. In addition, we discuss limitations and challenges of applications based on RNNs and future research directions. Finally, we summarize applications of RNNs in forecasting.

Comparing the performance of time series models with or without meteorological factors in predicting incident pulmonary tuberculosis in eastern China

“…The recurrent neural network (RNN) is a specific ANN with the ability to transfer information across time steps, as it can remember previous information and apply it to the current output calculation. The ability to model temporal dependencies makes it particularly appropriate to analyze a time series, which consists of a sequence of points that are not independent [ 9 , 10 ].…”

Real-Time Modulation of Physical Training Intensity Based on Wavelet Recursive Fuzzy Neural Networks

“…It was pointed out that their characteristics and the choice of basketball training intensity methods must be compatible with each other. e development of basic and special strength training is directed towards training in centrifugal and centripetal, positive, and negative forces, static to dynamic balance, power chains, and responsiveness, and the generation of power differs from the traditional speed and explosive power in muscle power, which is produced by neuromodulation of attack energy [7]. It introduces the concept of strength quality, illustration of body strength training in different parts, classification and description of characteristics according to body position and movement characteristics, intensity division, and flexibility training principles and methods to develop a good basis for training methods and overall training rules, which can be used as a basis and direction for the promotion of special strength of each sport; there are overall strength training methods according to the body power chain and training patterns [8].…”

CNSMolGen: A Bidirectional Recurrent Neural Network-Based Generative Model for De Novo Central Nervous System Drug Design

“…This is because RNNs can obtain high precision and good performance when processing time series predictions based on a large number of data sets. 22 In particular, Bjerrum and Threlfall introduced RNNs to generate chemically plausible and novel molecules by incorporating the long short-term memory (LSTM) network to overcome the challenges associated with vanishing gradient or explosion problems. 23 In addition, Wu et al proposed bidirectional long short-term memory (BiLSTM) attention network (BAN) to enhance the extraction of crucial features from the simplified molecular input line entry specification (SMILES) strings, leading to improved performance in molecular property prediction.…”

“…Subsequently, an array of different deep learning models has been developed in this field. − Among them, recurrent neural networks (RNNs) (Figure ) have shown remarkable performance. This is because RNNs can obtain high precision and good performance when processing time series predictions based on a large number of data sets . In particular, Bjerrum and Threlfall introduced RNNs to generate chemically plausible and novel molecules by incorporating the long short-term memory (LSTM) network to overcome the challenges associated with vanishing gradient or explosion problems .…”