LSTM Network
Long Short-Term Memory (LSTM) networks are a type of recurrent neural network designed to process sequential data, excelling at tasks requiring the retention of long-term dependencies. Current research focuses on improving LSTM performance through hybrid architectures (e.g., combining LSTMs with Convolutional Neural Networks or Transformers), optimizing model parameters, and addressing challenges like overfitting and computational cost in diverse applications. LSTMs' ability to model temporal patterns makes them valuable tools across numerous fields, including time series forecasting (weather, finance, energy), natural language processing, and signal processing, leading to advancements in areas like predictive maintenance and healthcare.
Papers
Comparative Analysis of LSTM, GRU, and Transformer Models for Stock Price Prediction
Jue Xiao, Tingting Deng, Shuochen Bi
TRIZ Method for Urban Building Energy Optimization: GWO-SARIMA-LSTM Forecasting model
Shirong Zheng, Shaobo Liu, Zhenhong Zhang, Dian Gu, Chunqiu Xia, Huadong Pang, Enock Mintah Ampaw