Research team
Expertise
I am a Vasileios Mygdals (male), born on 04/11/1986, Greece. I am a research associate specialised in Machine Learning and Computer Vision. I won a MSCA postdoctoral fellowship at the Marketing Research Group, University of Antwerp, for my current project titled CONVISE. I received a B.Sc. (2010) and a M.Sc. (2014) degrees in Biomedical Informatics, from the University of Central Greece and Aristotle University of Thessaloniki, respectively, and a Ph.D. in Informatics from the latter (2019). Since 2014 and during my PhD studies, I served at Aristotle University of Thessaloniki as a researcher and teaching assistant in the fields of machine learning, image processing, computer vision and pattern recognition, except a six-month period in 2017, where I served as a research assistant at the University of Bristol, United Kingdom. I have participated in 8 R&D collaborative projects funded by the European Union, executing research, integration, technical/financial management and administrative duties. I have (co-)authored more than 35 peer-reviewed papers in academic journals and international conferences. My current research interests include the areas of Attention-Based Marketing, Trustworthy AI, Adversarial Robustness, Computer Vision, Machine Learning, Robotic Perception. My publication record includes more than 35 peer-reviewed papers in academic journals (including the high-tier IEEE Transactions on Image Processing, Elsevier Pattern Recognition, Elsevier Neural Networks) and international conferences (e.g., IEEE International Conference on Image Processing, International Conference on Pattern Recognition), which attracted more than 530 citations, while my h-index is 12+ and i-10 index is 18+, according to Google Scholar. My research has produced theoretical methodological advances in general-purpose supervised classification/regression problems, including novel re-formulations of the optimisation process/criteria in order to account for additional information (e.g., geometric/semantic data properties such as the data distribution), task particularities, domain shifts, or encapsulate desirable model properties that need to simultaneously be adopted (e.g., privacy protection, increased inference speed, robustness). The devised optimisation problems have been applied to various shallow/deep model architectures and are typically easier-to-solve than their respective counterparts, leading to significant accuracy, precision, robustness, and speed gains, as they have been evaluated in a wide range of computer vision applications including object/face/human action recognition/detection/tracking and pose estimation.
Eye tracking for consumer attention sensing.
Abstract
This project is situated in the area of Attention-Based Marketing, which is the relevant emerging subdiscipline of Marketing, employing Consumer Attention (CA) sensing technology to understand and improve on (visual) marketing stimuli, marketing effectiveness, and consumer behaviours, by uncovering the fundamental interconnections between attention and action. The main objective of this research proposal is the evaluation of the influence of various different eye-tracking technologies in the overall eye-tracking experiment pipeline, in e-marketing settings. More precisely, it will study how do analysis results differ when the only difference in a variable that changes is the raw eye-tracking signal, find if there any particular settings including perhaps some pre/post-processing steps of the eye-tracking data or input data such that various eye-tracking technologies have exactly the same results in a particular marketing experiment and where do the observed differences converge when increasing the number of participants.Researcher(s)
- Promoter: Mygdalis Vasileios
Research team(s)
Project type(s)
- Research Project
Sensing, predicting and exploiting consumer visual attention in fast-paced marketing environments (CONVISE).
Abstract
In the attention economy, consumer attention is considered a finite, diminishing quantity acting as a currency that brands compete for to attract and maintain. Attention-Based Marketing is a relevant emerging subdiscipline of Marketing that studies consumer attention to understand and improve (visual) marketing stimuli, marketing effectiveness, and consumer behaviour by uncovering the fundamental interconnections between attention and action. Three main challenges in this area include how to: a) efficiently measure consumer attention in fast-paced environments (e.g., in electronic marketing), b) optimise marketing stimuli for attracting consumer attention and c) democratise consumer attention data (e.g., eye-sensing data/facial expressions) for improving the experience of individual consumers, while respecting their privacy at the same time. In-line with the above-mentioned challenges, the CONVISE project will integrate insights from the Marketing and Computer Vision disciplines to design methods for sensing, predicting, and exploiting consumer visual attention that can be used to optimise marketing efforts and enhance consumer well-being, in social media advertising settings. First, it will design video-based sensing technology for extracting reliable, market-relevant consumer visual attention maps, minimising the necessary human sensor (pre)-processing effort. Next, it will design tools for predicting consumer attention maps from the visual advertising content, without using any sensors, to produce objective advertisement evaluation metrics that can be employed during the visual advertising content design phase. Finally, it will experimentally evaluate the developed consumer attention prediction technology from a consumer behaviour standpoint and will develop sensing data de-identification technology, balancing between consumer privacy protection, perceived ad relevance and ad intrusiveness.Researcher(s)
- Promoter: Dens Nathalie
- Fellow: Mygdalis Vasileios
Research team(s)
Project website
Project type(s)
- Research Project