Concept Drift Detection and Adaptive Retraining of Malware Classification Models

One-Class Support Vector Machines (OCSVM) outperform Minibatch K-Means and Maximum Mean Discrepancy (MMD) in detecting statistical shifts in malware feature distributions. Testing across MLP, Random Forest, SVM, and XGBoost models shows OCSVM more effectively identifies the data distribution changes caused by evolving attacker techniques.