The application areas I worked and/or I am currently working in include emotion recognition from music, photovoltaic power generation under partial shading and applications in finance and adaptive computer memory controller design.
[June 2018] All good things come to an end. After 6 amazing years in the University of Manchester,
I am leaving at the end of the month. A sincere thanks to all my friends and colleagues here. I will miss you
all, but being away does not mean we won't be in touch!
Yet every end is a new beginning! From October 21st 2018 I will be joining the
Extrasolar Planets Group
of the Department of Physics and Astronomy of the
University College London.
I will be working with
Dr. Ingo Waldmann &
Prof. Giovanna Tinetti
and their team, using machine learning methods to sift through astronomical amounts of astronomical
data (got it?) and find patterns and irregularities that will help us better understand our universe. Sounds
exciting! I will provide a link to my new webpage in due time and this one will stop being updated.
[March 2018] Prof. Gavin Brown in his Inaugural Lecture made extensive mention to --among other outstanding pieces of work by fellow members of the team-- both our past work in Cost-Sensitive Boosting and our current work on applying machine learning on smart memory controllers. Congratulations to Gavin for his Professorship and to all members of the group for their amazing work. HAPPY to have been a part of it!
[September 2017] Attended ECML 2017 in Skopje, FYROM. I presented the paper "Gradient boosting models for photovoltaic power estimation under partial shading conditions"
at the International Workshop on Data Analytics for Renewable Energy Integration.
[August 2017] Attended the Data Science Summer School 2017 in Palaiseau, France, organized by the École Polytechnique. I presented a poster, titled, 'Better Boosting with Bandits'.
[July 2017] Attended Greek Stochastics iota in Milos, Greece. I contributed the talk 'Better Boosting with Bandits for
[July 2017] New publication: "Gradient boosting models for photovoltaic power estimation under partial shading conditions".
To be presented at the 5th International Workshop on Data Analytics for Renewable Energy Integration, held at ECML 2017.
[July 2017] Our paper "Cost-sensitive boosting algorithms: Do we really need them?",
published in Machine Learning Journal in September 2016 has been selected by the ACM as a
"Best of Computing, Notable Article of 2016". The articles are chosen through a nomination process
involving the editors-in-chief of ACM journals and senior academic / industry professionals,
to identify papers that are "the most interesting and influential that were published in 2016".
It has been downloaded over 2700 times so far.
[May 2017] I will be attending the Data Science Summer School organized by the École Polytechnique in Paris, France. The summer school takes place during August 28 - September 1 and I will also be presenting a poster there.
[April 2017] An i-python tutorial for the paper 'Cost-sensitive boosting algorithms: Do we really need them?'
can be found in GitHub. The users
can reproduce and significantly extend the experiments of the original paper. A more easy-to-use but less flexible version of
the code can also be found here. Any
feedback is welcome!
[December 2016] My EPSRC Doctoral Prize Fellowship has officially started! The project is titled 'Unifying Aspects of Machine Learning: From Boosting to Deep Learning'.
Its long term goal is to establish the underlying connections between boosting and deep learning and to leverage them to improve their theoretical understanding, scalability, predictive ability and applicability to diverse
scenarios (e.g. online learning, cost-sensitive learning, imbalanced class learning).
[December 2016] I have finished my PhD! My thesis, 'Cost-Sensitive Boosting: A Unified Approach', can be downloaded here.
[November 2016] My paper 'Cost-sensitive boosting algorithms: Do we really need them?' got the Best Paper Prize of the School of Computer Science at the University of Manchester 2016 (sponsored by IBM). Also, congratulations to my friend Konstantinos Sechidis for his Best Thesis Award!
[October 2016] A video of my ECML/PKDD 2016 plenary presentation is now available on youtube.
Lots of interesting questions from the audience. The third question was by none other than Zoubin Ghahramani!
[September 2016] Presented the paper 'Cost-sensitive boosting algorithms: Do we really need them?' at ECML/PKDD 2016 in Riva del Garda, Italy. The paper was selected for a plenary presentation as one of 12/129 eligible papers (top 9.3%).
[September 2016] I will be temporarily working as a Research Assistant for the EPSRC funded LAMBDA project (PI: Dr. Gavin Brown) until the end of November.
[September 2016] Submitted my PhD thesis titled 'Cost-sensitive Boosting: A Unified Approach'.
[June 2016] New publication: 'Cost-sensitive boosting algorithms: Do we really need them?', accepted in Machine Learning. To be presented in ECML/PKDD 2016 in Riva del Garda, Italy.
[November 2015] Gave the talk 'Cost-sensitive boosting algorithms: Do we really need them?' in the Research Symposium of the School of Computer Science of the University of Manchester.
[October 2015] Gave the lecture 'Cost-sensitive learning with AdaBoost' for the postgraduate course 'COMP61011: Foundations of Machine Learning' of the University of Manchester.
[July 2015] Attended Greek Stochastics eta in Chania, Crete, Greece. I presented a poster, titled, 'Predicting reliable probabilities with online Boosting'.
[July 2015] Presented the paper 'Calibrating AdaBoost for asymmetric learning' at MCS 2015 taking place at the Reisensburg Castle in Günzburg, Germany.
[June 2015] Attended the INIT/AERFAI 2015 summer school on machine learning at Benicŕssim, Valencia, Spain. Our poster, titled,
'Asymmetric boosting algorithms: Do we really need them?' received the best poster award. I also had the immense honour of
Prof. Janez Demšar, one of the invited speakers of the summer school, referring to my work on
various parts of his slides during his talk as a good example of how to compare methods empirically, but also on terms of simplicity, ease of use and
[May 2015] Gave the talk 'Optimal induction and its approximations' to the MLO group of the University of Manchester as part of the MLO Seminars.
[March 2015] New publication: 'Calibrating AdaBoost for asymmetric learning', accepted in MCS 2015.
[October 2014] Gave the lecture 'Introduction to AdaBoost' for the postgraduate course 'COMP61011: Foundations of Machine Learning' of the University of Manchester.
[September 2014] Attended ECML/PKDD 2014 in Nancy, France.
[August 2014] Presented 'Information theoretic feature selection in multi-label data through composite likelihood' in SSPR 2014 at Joensuu, Finland.
[April 2014] New publication: 'Information theoretic feature selection in multi-label data through composite likelihood', accepted in SSPR 2014.
[June 2013] Attended the INIT/AERFAI 2013 summer school on machine learning at Benicŕssim, Valencia, Spain.