Scholarlink

This research delves into the phenomenon of Cyanobacterial Harmful Algal Blooms (CyanoHABs), which are caused by cyanobacteria and pose significant threats to freshwater ecosystems such as lakes, rivers, reservoirs, and dams, due to the production of cyanotoxins that endanger human and animal health. The primary objective of this study is to protect public health, water resources and to improve efficient bloom management techniques using advanced Artificial Intelligence (AI) models to forecast the onset of CyanoHABs in freshwater ecosystems. Our approach employs a Generative Adversarial Network (GAN)-enhanced deep learning (DL) model in comparison with traditional regression methods such as Multi-Linear Regression (MLR), Support Vector Regression (SVR) refined through hyperparameter tuning, Decision Tree Regression (DTR), and Random Forest Regression (RFR). Our study demonstrates the effectiveness of GANs in synthesizing new data points that are diverse and realistic, thereby enriching existing datasets and enhancing the performance of DL models. Notably, the outstanding forecasting accuracy of the GAN-enhanced 1Dimensional Convolutional Neural Network (1D CNN) model, achieving an impressive R-squared (R2) value of 0.98, underscores its efficacy in predicting occurrences of CyanoHABs.

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

GAN-Enhanced Deep Learning Approach for Forecasting the Potentially Toxic Cyanobacteria in Dams

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Nour Djihane Amara,
  • Nadjette Dendani,
  • Amira Berrezzek,
  • Hana Khemissa,
  • Amel Saoudi,
  • Nabiha Azizi

摘要

This research delves into the phenomenon of Cyanobacterial Harmful Algal Blooms (CyanoHABs), which are caused by cyanobacteria and pose significant threats to freshwater ecosystems such as lakes, rivers, reservoirs, and dams, due to the production of cyanotoxins that endanger human and animal health. The primary objective of this study is to protect public health, water resources and to improve efficient bloom management techniques using advanced Artificial Intelligence (AI) models to forecast the onset of CyanoHABs in freshwater ecosystems. Our approach employs a Generative Adversarial Network (GAN)-enhanced deep learning (DL) model in comparison with traditional regression methods such as Multi-Linear Regression (MLR), Support Vector Regression (SVR) refined through hyperparameter tuning, Decision Tree Regression (DTR), and Random Forest Regression (RFR). Our study demonstrates the effectiveness of GANs in synthesizing new data points that are diverse and realistic, thereby enriching existing datasets and enhancing the performance of DL models. Notably, the outstanding forecasting accuracy of the GAN-enhanced 1Dimensional Convolutional Neural Network (1D CNN) model, achieving an impressive R-squared (R2) value of 0.98, underscores its efficacy in predicting occurrences of CyanoHABs.