Explainable and Analyst-Driven Random Forest for Intrusion Detection
Random Forest and other tree-ensemble classifiers achieve high accuracy in network intrusion detection; however, their aggregate decision logic prevents analysts from auditing or deploying individual predictions as operational rules. Post hoc explanation methods introduce latencies incompatible with security operation...