Abstract: Machine learning-based malware detection systems are vulnerable to adversarial attacks. Most existing attack methods require repeated queries, confidence score feedback, or knowledge of the ...