Hybrid Machine Learning Approach for Mosquito Species Classification Using Wingbeat Analysis

Abstract

Effective and precise techniques for mosquito species identification are required as mosquito-borne illnesses continue to pose serious threats to public health across the world. We provide a new hybrid machine-learning technique in this research work for the classification of mosquito species through the Wingbeat analysis. It analyzes the wingbeat of the mosquito species based on which it can identify the mosquito species. This method makes use of deep learning techniques. The hybrid technique attempts to provide robust and dependable classification performance by utilizing a wide range of machine learning methods, such as k-Nearest Neighbors (KNN), Random Forest, Multi-layer Perceptron (MLP), Support Vector Machines (SVM), and Gradient Boosting. To improve feature extraction and normalization, we apply a rigorous set of preprocessing techniques to a large dataset that includes wingbeat recordings from many mosquito species. By means of comprehensive testing and analysis, we prove that our method is effective in correctly detecting mosquito species, exhibiting better results than using separate machine learning algorithms. Our findings demonstrate how deep learning methods may support more conventional machine learning strategies in problems involving the categorization of mosquito species. We also address the implications of our results for ecological research and disease management initiatives, highlighting the significance of precise species identification in vector monitoring and epidemiological investigations.