The pixel takes it all
 
We are fascinated by how our brain interprets what we see. Emulation of our brain's ability to process visual signals has in turn inspired modern digital engines that analyse visual signals from cameras (such as pixels). A key component of these engines is the ability to understand the input visual data which is known as representation learning.   
 
In this lecture, I will give a historical overview of representation learning, from principal component analysis to the state-of-the-art of representation learning empowering all modern artificial intelligence (AI) applications. Along this timeline, I will share some of my work on data representations for various societal applications, from restoring paintings of Matisse to enhancing medical imaging diagnosis.  
 
The evolution of AI engines is epitomised by the picture accompanying this abstract which was made with Dalle-3 by Open AI. The picture's deficiencies will be used to motivate my most complex research undertaking to date. I will end with my vision for the future of AI and why pixels are the winners. (The lecture’s title is a play on the ABBA song "The winner takes it all".)


Professor Sotirios A. Tsaftaris, widely known as Sotos, is Chair in Machine Learning and Computer Vision at the University of Edinburgh. He also holds the Canon Medical/Royal Academy of Engineering Research Chair in Healthcare AI. He is the Director of the EPSRC-funded AI Hub for Causality in Healthcare AI with Real Data (CHAI). He also leads VIOS, a collaboration across industry and universities developing interdisciplinary AI. He is a Fellow of the European Laboratory for Learning and Intelligent Systems (ELLIS).
 
Since 2023 he is a visiting researcher with Archimedes RC a research centre of excellence in AI in Athens, Greece. Between 2016-2023 he was a Fellow of the Alan Turing Institute.
 
Before joining Edinburgh, Sotos was a faculty member of IMT Institute for Advanced Studies, Lucca, Italy and Northwestern University, USA.  
 
He has published extensively, particularly in interdisciplinary fields, with more than 200 journal and conference papers. His work has received several accolades, such as Best Paper Award (STACOM 2017, DART 2023, FAMI 2023, MIDL 2021), twice a Magna Cum Laude Award (ISMRM), a finalist for the Early Career Award (SCMR, 2011; SCMR, 2019 (Chartsias as PhD student)), and has appeared in journal covers and attract significant media coverage.
 
Sotos received MSc and PhD degrees from Northwestern University (USA) supported by a Murphy Fellowship and a scholarship from the Alexander S. Onassis Public Benefit Foundation. He also has an MEng from the Aristotle University of Thessaloniki (Greece).