I graduated from the Technion with a BSc in Electrical Engineering and BA in mathematics (both summa cum laude) in 1996. After that I spent almost four years as an intelligence officer with the Israeli Defense Forces. I was subsequently involved in a few ventures in the high-tech industry. I earned my PhD in Electrical Engineering from the Technion at 2002, under the supervision of Nahum Shimkin. I was then a Fulbright postdoctoral associate with LIDS (MIT) working with John Tsitsiklis for two years. I was at the Department of Electrical and Computer Engineering in McGill University from July 2004 until August 2010, where I held a Canada Research Chair in Machine Learning from 2005 to 2009. I have been with the Department of Electrical Engineering at the Technion since 2008 where I am a professor. I am father to Liam, Oliver and Dylan.
I am in the business of being a professor because I want to understand how to act and make decisions in dynamic, complex and uncertain environments. In plain language, I want to build machines (e.g., software agents) that learn, evolve, and improve over time. I work mostly in machine learning, but also in certain application domains.
- Reinforcement learning
- High dimensional statistics and learning
- Uncertainty and risk in decision making
- Learning and modeling dynamics from data
- Systems that include multiple decision makers: Multi-agent/distributed/many players/adaptive systems
See my Publications page for more details.
More specific research interests
- Machine Learning (theory, algorithms, and applications). High-dimensional problems with uncertainty in the data and modeling and learning dynamics (e.g., networks).
- Reinforcement Learning and Markov decision processes. Theory and application of Markov decision processes. I have worked quite a bit on adaptive control and learning algorithms for (large) stochastic systems in what is known as reinforcement learning.
- Learning, optimization and control under uncertainty. Robust and stochastic optimization and statistical analysis of such approaches.
- Games. Stochastic, dynamic, network, and differential games; applications in networks and resource sharing.
- Multi-agent systems. Especially learning in such systems (e.g., online learning and learning in games). The goal here is to design economic systems (e.g., markets) where equilibrium is also a good social outcome.
- Optimization of large scale problems. Especially combinatorial optimization using heuristic and statistical methods (e.g., the Cross Entropy method) and stochastic optimization.
- Power Grid. Especially in reliability, pricing, and decision making in large-scale power grids (smart grids). My approach is very much data-driven: I try to understand the actual dynamics of the grid so that I can propose concrete policies for control of the grid, as well as evaluate market mechanisms and anomalies. See, for example, the EU funded GARPUR project that looks at probabilistic reliability models for large-scale grids.
- Applications. I am interested and have worked (i.e., got to a semi-commercial prototype at least or plan to) on the following eclectic list of applications: large-scale communication network optimization, power management for laptops, adaptive compression of large databases, a learning agent for combat planes simulator, cognitive radio networks, human activity recognition and context identification on mobiles, stochastic approaches to decoding of LDPC codes.
- I am looking for a postdoc and a couple of graduate students to join my team. Please consider that working with me requires very strong mathematical skills and/or true hacking capabilities. Email me your resume and a brief explanation of what you want to do if you are interested.
Deep RL for Combinatorial Optimization.
*Joint supervision with Tamir Hazan
*Joint supervision with Shai Shen-Orr