Machine learning, sometimes called “AI,” is at the cutting edge of cybersecurity research. The fundamentals of machine learning are in statistics and knowledge of python is generally necessary to use machine learning. Thus, focus on Python and statistics first.
There are exciting research opportunities on campus exploring machine learning from several lenses. One is in the security of machine learning itself, known as adversarial machine learning, a field explored by Professor Dawn Song and the late Doug Tygar. For instance, how might attackers fool computer vision to cause, let’s say, a traffic accident? Another explored by Professor Raluca Ada Popa in the RISELab is secure collaborative learning, an approach that enables ML and knowledge discovery on data without transferring it to others.
The other approach, AI for security, is the subject of much hype and even companies deploying “Pseudo-AI”–companies that use AI as a marketing term but that rarely actually employ the technology. One notable exception is in anti-virus software, which has enough data to routinely use machine learning to spot variants of existing malware.