Those attending Black Hat and DEF CON in Las Vegas next week should also check out Security B-Sides (BSidesLV), where talks will range from threats against industrial control systems and mobile apps to how big data and deep learning can be used to mount a stronger defense. The event will also be heavy on talks about how to develop one’s career in the industry.
Two of Sophos’ data scientists will give talks at the event, held July 25 and 26 at Tuscany Suites. Other talks include “Something Wicked: Defensible Social Architecture in the context of Big Data, Behavioral Econ, Bot Hives, and Bad Actors,” by San Francisco-based security professional Allison Miller, and “Your Facts Are Not Safe with Us: Russian Information Operations as Social Engineering,” by Meagan Keim, a graduate student from the University of Maryland University College.
Sophos chief data scientist Joshua Saxe will present “The New Cat and Mouse Game: Attacking and Defending Machine Learning Based Software,” about ways the bad guys can manipulate machine learning to go on the attack. Saxe describes it this way in his talk description:
Machine learning is increasingly woven into software that determines what objects our cars recognize as obstacles, whether or not we have cancer, what news articles we should read, and whether or not we should have access to a building or device. Thus far, the technology community has focused on the benefits of machine learning rather than the security risks. And while the security community has raised concerns about machine learning, most security professionals aren’t also machine learning experts, and thus can miss ways in which machine learning systems can be manipulated.
My talk will help to close this gap, providing an overview of the kinds of attacks that are possible against machine learning systems, an overview of state-of-the-art methods for making machine learning systems more robust, and a live demonstration of the ways one can attack (and defend) a state-of-the-start machine learning based intrusion detection system.
Principal Sophos data scientist Richard Harang will present “Getting insight out of and back into deep neural networks.” He describes the talk this way:
Deep learning has emerged as a powerful tool for classifying malicious software artifacts, however the generic black-box nature of these classifiers makes it difficult to evaluate their results, diagnose model failures, or effectively incorporate existing knowledge into them. In particular, a single numerical output – either a binary label or a ‘maliciousness’ score – for some artifact doesn’t offer any insight as to what might be malicious about that artifact, or offer any starting point for further analysis. This is particularly important when examining such artifacts as malicious HTML pages, which often have small portions of malicious content distributed among much larger amounts of completely benign content.
In this applied talk, we present the LIME method developed by Ribeiro, Singh, and Guestrin, and show – with numerous demonstrations – how it can be adapted from the relatively straightforward domain of “explaining” text or image classifications to the much harder problem of supporting analysts in performing forensic analysis of malicious HTML documents. In particular, we can not only identify features of the document that are critical to performance of the model (as in the original work), but also use this approach to identify key components of the document that the model “thinks” are likely to contain malicious elements.
BSidesLV will also include a lockpick village, resume reviews and The New Hacker Pyramid, a contest that used to be presented at DEF CON but moved to BSidesLV a couple years ago.