Hello world! Since my youth, I have always been fascinated by computers and how they can leverage the tasks of our daily lives. Therefore, after secondary school, I enrolled at the Vrije Universiteit Brussel to study computer science. During these studies, I developed a growing interest in artificial intelligence and more precisely data mining and machine learning. Since I wanted to gain more insights in these techniques and the ideas behind the algorithms, I decided to pursue a PhD in artificial intelligence in 2011. During this doctoral study, I specialised myself in the domain of machine learning and reinforcement learning. I developed and proposed several reinforcement learning algorithms that are designated for multi-objective problem solving. These algorithms are presented at several international, peer-reviewed workshops and conferences. I plan to finish the PhD at the end of 2015. Thereafter, I would like to apply the knowledge and the insights that I have gathered in the industry.
Since October 2011
My research focusses on data mining, machine learning and optimisation techniques. More precisely, I investigate reinforcement learning algorithms. Reinforcement learning is a machine learning technique that learns through interacting with an environment and observing the feedback signal. By constantly building and refining estimates on the quality of the possible actions, the learning algorithm adjusts its hypothesis and acts accordingly. In my thesis, I extend reinforcement learning to problems with more than one feedback signal where the goal is to optimise in multiple dimensions at the same time. These newly developed algorithms are able to retrieve one or more Pareto optimal trade-off solutions that compromise different objectives simultaneously.
Machine learning researcher
January 2014 --- March 2014
During a 3-months working visit in the offices of Sensing & Control, I developed learning modules for the SCANERGY project. During my visit I investigated into a learning algorithm that can be incorporated in the smart meter of a household to provide a higher degree of balance between the supply and demand of electricity in the smart grid. The learning algorithm uses reinforcement learning to shift (part of) the consumption pattern of the household to time periods where the electricity price is low due to large quantities of injected (solar) energy. As a result, the energy price becomes more stable for all parties and outliers in the electricity pattern are reduced to a minimum.
Student job, website development
July 2009 --- August 2009
During this summer job, I created several webpages and linux scripts to automate information exchange at IMEC.
Doctor of Philosophy (Ph.D.) in Artificial Intelligence from Vrije Universiteit Brussel in 2015
Master of Science (MSc) in Computer Science from Vrije Universiteit Brussel in 2011
in Greek - Mathematics from Sint-Jozefscollege Woluwe in 2006
Training and Certification
Big Data Analysis with Revolution R Enterprise Certification
Data Analysis in R, the data.table Way Certification
Data Manipulation in R with dplyr Certification
Data Visualization in R with ggvis Certification
Intermediate R Certification
Intro to Statistics with R: Introduction Certification
Introduction to R Certification
Reporting with R Markdown Certification
Expert has 1 publications (Will be avalible with full profile)