Fertilizer prediction using serial exponential newton meta-heuristic algorithm-based convolutional neural network in IoT-based WSNs

• Published in: Computational biology and chemistry Vol. 119; p. 108561
• Main Authors: L T, Herlin, Lenin Fred, A.
• Format: Journal Article
• Language: English
• Published: England Elsevier Ltd 01.12.2025
• Subjects: deep learning; Fertilizer prediction; IoT; optimization algorithm; WSN; 3D CNN; CNN; DNN; MSVM-DAG-FFO; FNN; DL; KNN; optimization algorithm; ConvLSTM; GCEEC; IoT; BS; deep learning; MLR; RNN; SMOTE; LLT; SFSS; SNR; IoTSNA-CR; RCHS; SN; LDR; ML; ViT; NMA; LSTM; CH; SExpNMA; Fertilizer prediction; BiLSTM; SRLC; WSN; PH; 1D CNN
• ISSN: 1476-9271; 1476-928X; 1476-928X
• DOI: 10.1016/j.compbiolchem.2025.108561

This research presenteda Deep Learning (DL)for fertilizer prediction in IoT-based Wireless Sensor Network(WSN).At first, Serial Exponential Newton Meta-Heuristic Algorithm (SExpNMA) is used for Cluster Head (CH) routing and selection. SExpNMA is presented by integrating Newton Meta heuristic algorithm (NMA) and Serial Exponential Weighted Moving Average (SEWMA). Here, the routed data is handled using normalization and data cleaning, which is succeeded by data augmentation. Thereafter, feature fusion is donebasedoncorrelation of features combined with Bidirectional Long Short-Term Memory(BiLSTM).Thefertilizer prediction is done by using One-dimensional Convolutional neural network (1D CNN), which is trained bySExpNMAfor enhancing classifier parameters. By analyzing the outcomes of experiments SExpNMA based 1D CNN model has better performance when compared to conventional models with 94.2 % specificity,94.8 % sensitivity and 94.5 % accuracy. [Display omitted] •TTo perform feature fusion by a combination of correlation features and BiLSTM.•To develop an optimized DL mechanism, SExpNMA-based 1D-CNN for fertilizer prediction.•To evaluate proposed model metrics, accuracy, sensitivity, and specificity are used.