RESEARCH
Reinforcement Learning (RL): is a field of Machine Learning that provides a very general approach to Artificial Intelligence (AI). An agent (E.g. a robot) moves around in an environment by performing actions (E.g. move left/right) based on its current state (E.g. x,y location) in order to complete a task. Whenever the agent performs an action, the agent recieves a reward. This reward is positive if it leads the agent to a desirable state or negative if the agent reaches an undesirable one. The goal of the agent is to complete the task such that it maximizes the rewards that it accumulates during the task.
Real-world RL: I am passionate about using RL algorithms to solve Google-scale real-world applications. Previously in a paper that I co-authored, we identified 9 challenges that, if solved, will enable RL to scale to solving real-world problems. These challenges are based on our collective experience working on applied RL problems. The challenges are:
​
1. Training off-line from the fixed logs of an external behavior policy.
2. Learning on the real system from limited samples.
3. High-dimensional continuous state and action spaces.
4. Safety constraints that should never or at least rarely be violated.
5. Tasks that may be partially observable, alternatively viewed as non-stationary or stochastic.
6. Reward functions that are unspecified, multi-objective,or risk-sensitive.
7. System operators who desire explainable policies and actions.
8. Inference that must happen in real-time at the control frequency of the system.
9. Large and/or unknown delays in the system actuators,sensors, or rewards.
​
My research currently focuses on solving these problems.
Computer Vision and Robotics: are also active interests of mine. Although this is not my main research, I have been actively involved in developing object recognition algorithms for the DARPA Robotics challenge as a representative of the Israeli Team Robil. I have also developed a highly competitive Visual Odometry algorithm called CFORB. In addition, I was a member of Team Edinferno at the 2013 Robocup in the Standard Platform League, which was held in Eindhoven.
