Stephanie cuthbertson - director, android - google google i/0 2019 keynote speech. Palo alto networks, wildfire malware analysis service. SHIELD exhibits state-of-the-art performance for traditional malware detection, with an accuracy of 99.52%. Existing anti-malware programs fail to detect obfuscated backdoor, while SHIELD successfully flagged the obfuscated backdoor as a malicious application. Further, we created 500 obfuscated backdoor applications to evaluate the effectiveness of SHIELD with respect to other existing mobile anti-malware programs. We validate our approach of unseen malware detection using the CICandMal2020 and AMD benchmarks datasets while achieving detection rates of 94% and 87%, respectively. SHIELD uses multimodal autoencoder (MAE) technique, which cuts down the dependency on feature engineering and automatically discovers the relevant features for malware detection. In this paper, we present a multimodal deep learning framework, for unseen Android malware detection, called SHIELD, which employs Markov image of opcodes and dynamic APIs. Traditional malware detection methods are ineffective as Android malware use sophisticated obfuscation and adapt to the anti-virus defenses. Consequently, the Android OS continues to be a prime target for serious malware attacks. The widespread adoption of Android OS in recent years is due to its openness and flexibility.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |