Samantaray (2026) Hybrid a Symmetric Huber Loss Function-Based ELM Approach for Average Temperature Prediction: A Case Study on Kokernag, Jhelum River Basin, India

Identification

• Journal: Lecture notes in networks and systems
• Year: 2026
• Date: 2026-01-01
• Authors: Sandeep Samantaray
• DOI: 10.1007/978-981-95-2872-1_44

Research Groups

• Department of Civil Engineering, NIT Srinagar, Jammu and Kashmir, India
• Water Resources Management Center, NIT Srinagar, Jammu and Kashmir, India

Short Summary

This study introduces a novel Hybrid Asymmetric Huber Loss Function-Based Extreme Learning Machine (AHELM) model for accurate average temperature prediction, demonstrating superior performance over traditional ELM in the Jhelum River Basin, India.

Objective

• To develop and evaluate a novel Hybrid Asymmetric Huber Loss Function-Based Extreme Learning Machine (AHELM) model to improve average temperature prediction accuracy, particularly in handling outliers.

Study Configuration

• Spatial Scale: Kokernag station, Jhelum River Basin, India
• Temporal Scale: Time-series analysis of historical temperature data using time-lagged inputs (ATt-1, ATt-2, ATt-3, ATt-4).

Methodology and Data

• Models used: Hybrid Asymmetric Huber Loss Function-Based Extreme Learning Machine (AHELM), Extreme Learning Machine (ELM)
• Data sources: Historical temperature data from Kokernag station

Main Results

• The proposed AHELM model significantly outperformed the standalone ELM model.
• AHELM achieved a coefficient of determination (R²) of 0.9812.
• AHELM achieved a Nash-Sutcliffe Efficiency (NSE) of 0.9756.
• The hybrid model demonstrated superior accuracy and enhanced reliability in temperature forecasting.

Contributions

• Introduction of a novel hybrid machine learning model (AHELM) specifically designed for average temperature prediction.
• Integration of an Asymmetric Huber Loss Function into an ELM model to effectively handle outliers and improve prediction accuracy.
• Demonstration of superior predictive performance and enhanced forecasting reliability compared to traditional ELM models.

Funding

• Not specified in the provided text.

Citation

@article{Samantaray2026Hybrid,
  author = {Samantaray, Sandeep},
  title = {Hybrid a Symmetric Huber Loss Function-Based ELM Approach for Average Temperature Prediction: A Case Study on Kokernag, Jhelum River Basin, India},
  journal = {Lecture notes in networks and systems},
  year = {2026},
  doi = {10.1007/978-981-95-2872-1_44},
  url = {https://doi.org/10.1007/978-981-95-2872-1_44}
}