Scholarlink

Modern agriculture is under growing pressure from problems like population growth-driven rising food demand, water scarcity, and climate change. Artificial intelligence (AI) methods, especially deep learning for examining varied data and attaining high yields, are being depended upon to meet these obstacles. Using deep learning methods, this paper emphasises crop monitoring and yield forecasting. Using long short-term memory (LSTM) networks, temporal data was feature-extracted. Principal Component Analysis (PCA) was used to reduce dimensionality following feature extraction, three models were then trained using AdaBoost, LightGBM as an ensemble decision making, and Random Forest as a non-linear model, which were then optimized using the Bayesian optimization (BO) algorithm. The findings indicated that LSTM outperformed all tests by 100%, hence proving the use of the extracted characteristics. These results indicate that LSTM technique is effective in extract accurate features that contribute to increasing crop productivity, opening up broad prospects for data-driven smart agriculture applications.

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

Data-Driven Agriculture Using LSTM and PCA: Enhancing Crop Yield with Ensemble and Bayesian Learning Techniques

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Mohammed Khalaf Mahmood,
  • Wisam Dawood Abdullah

摘要

Modern agriculture is under growing pressure from problems like population growth-driven rising food demand, water scarcity, and climate change. Artificial intelligence (AI) methods, especially deep learning for examining varied data and attaining high yields, are being depended upon to meet these obstacles. Using deep learning methods, this paper emphasises crop monitoring and yield forecasting. Using long short-term memory (LSTM) networks, temporal data was feature-extracted. Principal Component Analysis (PCA) was used to reduce dimensionality following feature extraction, three models were then trained using AdaBoost, LightGBM as an ensemble decision making, and Random Forest as a non-linear model, which were then optimized using the Bayesian optimization (BO) algorithm. The findings indicated that LSTM outperformed all tests by 100%, hence proving the use of the extracted characteristics. These results indicate that LSTM technique is effective in extract accurate features that contribute to increasing crop productivity, opening up broad prospects for data-driven smart agriculture applications.