From Music to Machine Learning

Prof. Bracha Shapira had dreamed of becoming a concert pianist, and before focusing on artificial intelligence and computational learning, she completed a bachelor's degree in musicology. She studied computer science to earn a living, and after completing another bachelor's degree, she worked as a programmer. At the same time, she studied for a master's degree; "for the fun of it" in her own words. After a few years of working in the hi-tech industry, Shapira returned to academia, where she has been ever since. She moved from computer science to engineering, and this combination helped her diverse research agenda. She studied musicology and computer science at Bar-Ilan University; her master's degree in computer science is from the Hebrew University of Jerusalem, and she completed her doctoral dissertation at Ben-Gurion University, where she arrived in 1995. "I was the University’s first doctoral student in information systems engineering," she recalled. For her postdoc, Shapira went to Rutgers University in the USA.

In her doctoral research, Shapira developed a model for the automatic characterization of user groups for the purpose of intelligent e-mail filtering. Later, she collaborated in a study that automatically identified groups of internet users, and in particular groups of terrorists, based on their surfing behavior. Then, together with BGU’s Prof. Yuval Elovici, she developed a method to protect users from tracking their online behavior. Today, she studies different methods of machine learning: using existing or acquired information to "teach" a computer to solve, and teach itself to improve in solving new problems.

Between 2011 and 2017, Prof. Shapira chaired the Department of Software & Information Systems Engineering, and later was the Deputy Dean of the Faculty of Engineering Sciences. She currently heads the Academic Committee of the Council for the Advancement of Women at the Ministry of Innovation, Science and Technology.

Shapira has played a key role in BGU’s transformation into a national leader in cyber security technological research, with an emphasis on the application of methods from machine learning. She has worked on the detection of anomalies (unusual behaviors and hostile codes) using sophisticated methods, on the simulation of cyberattacks and predicting the unfolding of such attacks, as well as planning effective network defenses and identifying threats in social networks.

Can you explain how social networks are exposed to malicious activity?

"Users, especially the young, are not really aware of the risks of revealing their information online. Social networks are fertile grounds for spying on people and companies. You can learn about a person's habits via her friends or family, without that person publishing any information herself; there are many ways to discover personal information (for example through a fictitious friendship offer). Social networks are infested with sex offenders looking for unwary users. But, in this context, I would like to mention that we have developed methods of identifying bots and alert users."

Adapted from issue 136  of Aleph-Bet-Gimmel, the University’s Hebrew language magazine. For the original article.