Anestis Fachantidis
Research Interests
The motivations were always there... Revealing the deep structure underlying raw data, extracting "hidden" rules guiding a behaviour or a policy, developing systems that adapt and behave based on learning and not on hard-coded rules, designing and implementing intelligently amplified systems that could assist humans achieve their goals. These early motivations of mine are still here, driving my research in Reinforcement Learning and Machine Learning.
Currently I’m working on Transfer Learning for Reinforcement Learning, that is, transferring an agent's knowledge for a specific task to another relevant but more complex task, so that he can utilize this knowledge and not having to learn from scratch.
Moreover I am very interested on novel (real life) applications of Machine Learning in many, vastly different domains, such as, computational advertisement, systems for Green ICT, Sensor Networks, RL and Robotics.