- George Magoulas
- London, United Kingdom
- I conduct interdisciplinary research in software environments that exhibit different levels of intelligence and are capable of learning from data. My latest projects are in deep learning for psychophysiological data modelling and classification, and in intelligent learning environments. In these projects, I lead research on the development of intelligent components, which employ machine learning, sometimes combined with knowledge engineering methods, and the design and implementation of personalisation technologies. I was educated at the School of Engineering, University of Patras, Greece (BEng/MEng, Dr. Eng), and hold a PGCE from Brunel University, UK. Before joining academia I held R&D positions in the cement and automotive industries where I worked on the development of embedded systems employing soft computing and machine learning methodologies. My research received best paper awards from the IEEE (2000 and 2008), the European Network on Intelligent Technologies for Smart Adaptive Systems (2001 and 2004), the International Association for Development of the Information Society (2006), the Association for Computing Machinery (2009) and KES International (2010).
10 Jan 2017
Training Convolutional Networks with Weight–wise Adaptive Learning Rates at the 25th ESANN 2017.
Design and evaluation of a case-based system for modelling exploratory learning behaviour of math generalisation, in IEEE TLT
Deep Learning Parkinson’s from Smartphone Data at the IEEE PerCom 2017.
3 Jan 2017
My group received an NVIDIA Grant to support research on Deep Learning Ensembles. In this project, Alan Mosca will investigate novel methods for efficient creation and training of deep learning ensembles, and will develop tools for parallel training of deep learning ensembles on multiple GPUs.
8 Oct 2016
Deep Incremental Boosting at GCAI 2016.
Learning Input Features Representations in Deep Learning
Regularizing Deep Learning Ensembles by Distillation at ECAI's CIMA 2016.
Bounding the Search Space for Global Optimization of Neural Networks Learning Error: An Interval Analysis Approach in JMLR.
A Review, Timeline, and Categorization of Learning Design Tools at ICWL 2016; shortlisted for the ICWL 2016 best paper award.
Approaches to Design for Learning at ICWL 2016.
An Architecture for Smart Lifelong Learning Design at ICSLE 2016.
Ubiquitous Learning Architecture to Enable Learning Path Design across the Cumulative Learning Continuum, in Informatics.
25 May 2014
I was keynote speaker at the 8th Hellenic Conference on Artificial Intelligence (SETN'14), which was held in Ioannina, Greece, in May 2014. My talk entitled "Making educational software more reactive to users: how AI can help" presented externally-funded research on the design and development of intelligent components for interactive learning environments. At the same meeting, I received honorary member status of the Hellenic Artificial Intelligence Society .