Robusztus malware detekció gépi tanulással (MILAB)

We aim to design machine learning solutions to detect malwares on resource-constrained embedded (e.g., IoT) devices in a robust and privacy-preserving way. Our specific objectives are: (1) develop malware detection tools that are scalable to resource constrained embedded devices, (2) mitigate evasion attacks (adversarial examples) against such tools, (3) detect and mitigate unintended information leakage about the potentially sensitive malware database (training data).