Adversarial Evasion in Non-Stationary Malware Detection: Minimizing Drift Signals through Similarity-Constrained Perturbations

ArXi:2604.21310v1 Announce Type: cross Deep learning has emerged as a powerful approach for malware detection, nstrating impressive accuracy across various data representations. However, these models face critical limitations in real-world, non-stationary environments where both malware characteristics and detection systems continuously evolve.