Pleased to meet you: Dr. Ilan Reuven Cohen

If you ever wondered why Facebook or your favorite news site offers certain ads (a four-bedroom apartment in Rishon Lezion, teeth whitening in Yavne, divorce attorneys in Haifa), the answer lies in algorithmic game theory. “Allocating ads to users is an algorithmic problem. When you have a certain user, several companies want to catch their attention with ads, and the algorithm’s task is to decide which company is going to show up in order to maximize profits,” explains Dr. Ilan Reuven Cohen. “It’s a problem that’s worth a lot of money: the online ad business makes tens of billions of dollars per year; it is constantly growing, and it involves mega-corporations such as Facebook and Google.”

Algorithmic game theory is one of Dr. Cohen’s (38) areas of research. Just this year, Cohen joined the Faculty of Engineering as a faculty member in the Faculty Engineering and Information Systems program. He also researches online and stochastic algorithms, operations research, and optimization, particularly factors pertaining to scheduling. “In the internet age, a significant portion of our work has turned to cloud computing, where tasks are sent to huge, remote server farms. As a result, plenty of issues arose concerning the optimization of servers used within the farms. Their energy consumption is immense, and any improvement is significant, both in terms of energy consumption and costs,” he explains. “At the same time, because the work is done via cloud computing, on remote computers, it comes with a lot of uncertainty: when will the task come in, how many, how long would each task take to execute, which resources each task would need, etc. So must of my research focuses on this aspect, the uncertainty of these servers, in an attempt to improve their performance and efficiency.”

Dr. Cohen started out in mapping algorithms. “After acquiring a BSc in computer science at the Technion, I served in the IDF’s Intelligence Corps, developing and modeling algorithms for the prediction and mapping unit,” he shares. “During my service, I got my MSc from Tel Aviv University, and once I completed my service I went on to acquire my PhD with Prof. Yossi Azar. At the same time, I helped found a startup in the field of mapping.” He then went on to pursue a post-doctorate at UC Berkeley’s Simons Institute, at Carnegie Mellon in Pittsburg, and at Centrum Wiskunda & Informatica in Amsterdam, researching algorithms in states of uncertainty.

“The classic computer science model assumes that when you want to solve a problem, you know nothing about it beforehand – but once you have the problem in front of you, you have all the data,” he explains. “The problem is that in many cases, some conditions are not known in advance. Navigation, for example; say an app like WAZE wants to take me from point A to point B. in the classic model, there’s some sort of road system where distances are known, and everything it has to do is use it to calculate the best route. But in navigation, we have a lot of unforeseeable elements such as congestion or accidents. So the question is how, in terms of uncertainty, we can plan the algorithm to produce the best possible outcome.”

Cohen also researches social welfare improvement, again in the aspect of algorithmic game theory. One such issue is parking problems, and you might not like his conclusions: “the question we faced was whether city planning should allow free parking or charge for parking, and our study has shown that if you allegedly let things happen naturally – anyone can park anywhere – social welfare is harmed,” he says. “In the free parking model, each driver parks at the spot closest to their destination, but if we look at it in a larger perspective, that spot might be better suited for someone else’s needs who now have to park farther. We proposed a solution of adding a monetary incentive – in other words, charging different rates for parking in different locations. In essence, we created an excuse for charging for parking, but backed it up with theoretical proof that it would improve social welfare: people would park closer to their destination, and everyone’s quality of life would improve.”

Starting next year, Dr. Cohen will teach courses pertaining to optimization and operations research. He is currently looking for master’s students to join his group. If you have an interest in theoretical and practical research in the field of operations research and algorithmic game theory, contact him at ilan-reuven.cohen@biu.ac.il.