Security of Machine Learning - Stian Kristoffersen

Stian Kristoffersen from Deepinsight will come and talk about security of machine learning. The presentation will be held in English.

Machine Learning (ML) continues to be a trendy tool for many types of problems, including security. But is ML itself secure? This talk will give an introduction to attacks on ML like changing predictions, recovering sensitive information, and stealing someone else's models. Examples include misdirection by changing a few pixels in an image, by using 3D printed models, and by hiding long messages in short sound bites. We will conclude with some current research directions to mitigate these attacks. Prior experience with ML is useful, but not required.

Pizza and soda will be served at the meetup, sponsored by Deepinsight (https://deepinsight.io/).

A big thanks to mnemonic (https://www.mnemonic.no/) for supporting the OWASP Norway Day 2018 as a platinum sponsor.