ML

Securing ML-based Android Malware Detectors: A Defensive Feature Selection Approach against Backdoor Attacks
Bartłomiej Marek, Kacper Pieniążek, Filip Ratajczak, Wojciech Adamczyk, Bartosz Bok, Mateusz Krzysztoń
Jul 1, 2024
Multilingual aspects of audio DeepFakes
Comparison of state-of-the-art deep learning models for audio deepfakes detection in the context of audio DeepFakes across 8 languages (English, German, Polish, Ukrainian, Russian, Spanish, French, Italian).
Verification and identification behavioural biometric system based on typing errors in English in Python.
The system joins keylogger and text editor functionalities to gather data. Features describing each user are extracted using natural language processing techniques and metrics.
Securing ML-based Android Malware Detectors: A Defensive Feature Selection Approach against Backdoor Attacks
Proposal of an effective feature selection method to reduce the vulnerability of a machine learning system to a backdoor attack.