Scholarlink

<p>In this research, the big data-based rainfall prediction is done using the Hybrid Long Short-Term Memory Fused Deep Stacked Auto Encoder Network (LSAE-Net). The process starts by partitioning the time series data into numerous parts using the Bayesian Fuzzy clustering (BFC) method, which is then transferred into the MapReduce framework with mappers and reducers. The transferred data allowed into various mappers are subjected to the extraction of technical indicators, and the Deep Residual Network (DRN) is used for the feature fusion process. The reducer phase acquires all the features collected from all the mappers and merge them. Finally, the merged feature is allowed to the proposed LSAE-Net. Here, LSAE-Net is formed by combining the Deep-stacked Auto Encoder Network (DSAE) and Long Short-Term Memory (LSTM) for prediction of rainfall. Furthermore, the efficiency of LSAE-Net is found by various metrics, like Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Square (RMSE), Coefficient of Determination (R2 score), and Nash–Sutcliffe Efficiency (NSE), which attained superior values of 0.074, 0.035, 0.188, 0.964, and 0.954, respectively.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Optimizing Rainfall Prediction in Agriculture with Deep Learning: a Hybrid LSTM and Deep Stacked Autoencoder Network for Big Data

  • 期刊论文

  • 发表日期：

    2026年2月16日

查看全文
  • Vinoth R,
  • Gangadevi G,
  • C. G. Balaji,
  • Cypto J

摘要

In this research, the big data-based rainfall prediction is done using the Hybrid Long Short-Term Memory Fused Deep Stacked Auto Encoder Network (LSAE-Net). The process starts by partitioning the time series data into numerous parts using the Bayesian Fuzzy clustering (BFC) method, which is then transferred into the MapReduce framework with mappers and reducers. The transferred data allowed into various mappers are subjected to the extraction of technical indicators, and the Deep Residual Network (DRN) is used for the feature fusion process. The reducer phase acquires all the features collected from all the mappers and merge them. Finally, the merged feature is allowed to the proposed LSAE-Net. Here, LSAE-Net is formed by combining the Deep-stacked Auto Encoder Network (DSAE) and Long Short-Term Memory (LSTM) for prediction of rainfall. Furthermore, the efficiency of LSAE-Net is found by various metrics, like Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Square (RMSE), Coefficient of Determination (R2 score), and Nash–Sutcliffe Efficiency (NSE), which attained superior values of 0.074, 0.035, 0.188, 0.964, and 0.954, respectively.