The paper introduces the Bitterling Fish Optimization (BFO) algorithm, a meta-heuristic optimization method inspired by the mating behavior of bitterling fish. The algorithm is designed to solve optimization problems and is compared with other meta-heuristic algorithms such as Gray Wolf Optimization, Whale Optimization, Butterfly Optimization, Harris Hawks Optimization, and Black Widow Optimization. The BFO algorithm is found to be more accurate in various benchmark functions and practical applications. The authors present a discrete binary version of the BFO algorithm and apply it to feature selection in the UNSW-NB15 dataset, achieving an accuracy of 96.72%. They also integrate the BFO algorithm with an artificial neural network (ANN) to detect network intrusions, achieving an accuracy of 99.14%, precision of 98.87%, and sensitivity of 98.85% on the NSL KDD dataset. The proposed method outperforms other machine learning approaches like NNIA, DT, RF, XGBoost, and CNN in intrusion detection. Additionally, the BFO algorithm is used to improve K-means clustering on datasets related to COVID-19, diabetes, and kidney disease, showing better performance compared to iECA*, ECA*, and GENCLUST++. The paper also reviews related works and discusses the advantages and challenges of various meta-heuristic algorithms. The BFO algorithm is evaluated through benchmark functions, engineering optimization problems, network traffic classification, network intrusion detection, phishing attack detection, and disease detection. The algorithm's effectiveness is demonstrated through detailed analyses, including convergence, stability, and runtime comparisons with other meta-heuristic algorithms.The paper introduces the Bitterling Fish Optimization (BFO) algorithm, a meta-heuristic optimization method inspired by the mating behavior of bitterling fish. The algorithm is designed to solve optimization problems and is compared with other meta-heuristic algorithms such as Gray Wolf Optimization, Whale Optimization, Butterfly Optimization, Harris Hawks Optimization, and Black Widow Optimization. The BFO algorithm is found to be more accurate in various benchmark functions and practical applications. The authors present a discrete binary version of the BFO algorithm and apply it to feature selection in the UNSW-NB15 dataset, achieving an accuracy of 96.72%. They also integrate the BFO algorithm with an artificial neural network (ANN) to detect network intrusions, achieving an accuracy of 99.14%, precision of 98.87%, and sensitivity of 98.85% on the NSL KDD dataset. The proposed method outperforms other machine learning approaches like NNIA, DT, RF, XGBoost, and CNN in intrusion detection. Additionally, the BFO algorithm is used to improve K-means clustering on datasets related to COVID-19, diabetes, and kidney disease, showing better performance compared to iECA*, ECA*, and GENCLUST++. The paper also reviews related works and discusses the advantages and challenges of various meta-heuristic algorithms. The BFO algorithm is evaluated through benchmark functions, engineering optimization problems, network traffic classification, network intrusion detection, phishing attack detection, and disease detection. The algorithm's effectiveness is demonstrated through detailed analyses, including convergence, stability, and runtime comparisons with other meta-heuristic algorithms.