One Class Support Vector Machine

One-class Support Vector Machines (OC-SVMs) are machine learning algorithms designed for anomaly detection using only data from the normal class. Current research focuses on improving OC-SVM efficiency for large datasets through techniques like variable subsampling, randomized measurements, and integrating with dictionary learning or autoencoders to enhance feature extraction and model robustness. These advancements aim to address computational limitations and improve accuracy in various applications, including anomaly detection in network logs, medical imaging, and structural health monitoring. The resulting models show promise for improving the speed and accuracy of anomaly detection across diverse fields.