Related Experiment Videos
Finite-Element-Informed Pyramid Neural Network With Draco Lizard Optimizer for Accurate and Efficient Water
Navala Rani Gunti1, Kodukula Subrahmanyam1
1Department of Computer Science and Engineering, Koneru Lakshmaiah Education Foundation, Guntur, Andhra Pradesh, India.
Summary
A new Finite-Element-Informed Pyramid Neural Network with Draco Lizard Optimizer (F-E-IPNNet-DLO) effectively classifies water contamination. This model offers high accuracy for sustainable aquaculture and environmental protection.
Area of Science:
- Environmental Science
- Aquaculture Technology
- Artificial Intelligence
Background:
- Water contamination poses a significant threat to aquaculture sustainability and aquatic health.
- Conventional methods for assessing water quality are often slow, costly, and error-prone.
- Existing deep learning approaches face challenges with data variability, interpretability, and predictive accuracy.
Purpose of the Study:
- To introduce a novel, robust classifier for water contamination detection in aquaculture.
- To address the limitations of conventional and current deep learning methods in water quality assessment.
- To enhance the efficiency and reliability of monitoring pollution clean-up in aquatic ecosystems.
Main Methods:
- A Finite-Element-Informed Pyramid Neural Network with Draco Lizard Optimizer (F-E-IPNNet-DLO) was developed.
- Data preprocessing involved Trimmed Scores Regression to K-Means Clustering (TSRK-MC) for noise and outlier management.
- Feature prediction and extraction were performed using the Billiards-inspired Ebola Search Optimization Algorithm (B-ESOA) and a pyramid neural network, multi-scaled by the Draco Lizard Optimizer (DLO).
Main Results:
- The F-E-IPNNet-DLO model achieved exceptional classification performance, with accuracy, recall, precision, F1-score, and specificity exceeding 99.96% on two benchmark datasets.
- The model demonstrated minimal predictive errors, highlighting its reliability.
- The preprocessing and feature selection methods effectively handled data noise, outliers, and missing values.
Conclusions:
- The proposed F-E-IPNNet-DLO model is a highly reliable and efficient tool for water contamination classification.
- This approach significantly advances the capabilities for sustainable aquaculture and effective environmental conservation efforts.
- The model's high performance validates its applicability in real-world water quality monitoring scenarios.