A Whirlwind Tour of Academic Techniques for Real-World Security Researchers

No ratings

Presented at Breakpoint 2013 by

Silvio Cesare

This talk will discuss some of the academic tools and techniques discovered and implemented whilst completing a Doctorate degree that are useful to real-world security researchers. Each of the techniques discussed will include real-world usage. Machine learning brings classification and clustering that implements supervised and unsupervised learning. You can use this to classify and cluster malware, emails, twitter messages, web pages, and the list goes on. You can even use classification to help find vulnerabilities. You'll learn that in less time than an afternoon you can implement and evaluate these techniques on your data sets using free software. Well that's cool, but what then? You probably all know about abstract data types and how they are useful to programmers, but what about abstractions used by mathematicians and computer scientists? Using vectors, sets, and graphs, you'll learn how these abstractions can be compared, approximated, indexed, and searched that will let you find near duplicates from your data set which could be things such as log files, network packets, stack traces, web pages, network topologies, or even malware. Finally, we'll look at the academic world of program analysis. You'll learn about monotone frameworks, abstract interpretation, data-flow analysis, white-box fuzzing, and what those SMT solvers are all about. Hopefully, by the time this talk is over, you'll see that academia offers tools that are useful right now for many of us in the real-world.