Malign Machine Learning Models and bad DICOM

Zeronighs 2019 AISec releases: how to insert malware into TensorFlow and PyTorch models and hack NVIDIA Clara ML pipeline with DICOM image.





MALIGN MACHINE LEARNING MODELS

ROMAN PALKIN

Backed up with real examples, this talk reviews the capabilities of widely-used frameworks TensorFlow and PyTorch for creating and spreading malicious software as well as implementing covert data communication channels. The purpose of this presentation is to draw attention of the community to the danger posed by careless use of Machine Learning models from unreliable sources.

HACKING MEDICAL IMAGING WITH DICOM

MARIA NEDYAK

Currently, DICOM is one of the core technologies used in Machine Learning Medical Imaging, which is growing more popular and relevant. The subject of this talk is discovery of vulnerabilities in a widely-used application for DICOM. We’ll also discuss the vulnerabilities discovered during the security assessment.

Enjoy