Machine Learning for Medicine and Healthcare
Mihaela van der Schaar, University of Cambridge, United Kingdom
Everything Has Been Done, It Is Our Job to Do It One Better in Image Matching and Localisation
Krystian Mikolajczyk, Imperial College London, United Kingdom
Keep on Learning
Tinne Tuytelaars, Electrical Engineering, KU Leuven, Belgium
Image and Video Generation: A Deep Learning Approach
Nicu Sebe, University of Trento, Italy
Machine Learning for Medicine and Healthcare
Mihaela van der Schaar
University of Cambridge
United Kingdom
![]()
Brief Bio
Mihaela van der Schaar is the John Humphrey Plummer Professor of Machine Learning, Artificial Intelligence and Medicine at the University of Cambridge, a Fellow at The Alan Turing Institute in London, and a Chancellor’s Professor at UCLA. In addition to leading the van der Schaar Lab, Mihaela is founder and director of the Cambridge Centre for AI in Medicine (CCAIM).
Mihaela was elected IEEE Fellow in 2009. She has received numerous awards, including the Oon Prize on Preventative Medicine from the University of Cambridge (2018), a National Science Foundation CAREER Award (2004), 3 IBM Faculty Awards, the IBM Exploratory Stream Analytics Innovation Award, the Philips Make a Difference Award and several best paper awards, including the IEEE Darlington Award.
Mihaela is personally credited as inventor on 35 USA patents (the majority of which are listed here), many of which are still frequently cited and adopted in standards. She has made over 45 contributions to international standards for which she received 3 ISO Awards.
Abstract
Medicine stands apart from other areas where machine learning can be applied. While we have seen advances in other fields with lots of data, it is not the volume of data that makes medicine so hard, it is the challenges arising from extracting actionable information from the complexity of the data. It is these challenges that make medicine the most exciting area for anyone who is really interested in the frontiers of machine learning – giving us real-world problems where the solutions are ones that are societally important and which potentially impact on us all. Think Covid 19!
In this talk I will show how machine learning is transforming medicine and how medicine is driving new advances in machine learning, including new methodologies in time-series, causal inference, interpretable and explainable machine learning, as well as the development of new machine learning areas - quantitative epistemology.
Everything Has Been Done, It Is Our Job to Do It One Better in Image Matching and Localisation
Krystian Mikolajczyk
Imperial College London
United Kingdom
Brief Bio
Krystian Mikolajczyk is a professor at Imperial College London. He did his undergraduate study at the University of Science and Technology (AGH) in Krakow, Poland, and completed his PhD degree at the Institute National Polytechnique de Grenoble, France. He then worked as a research assistant in INRIA, University of Oxford and Technical University of Darmstadt (Germany), before joining the University of Surrey, and Imperial College London in 2015. His main area of expertise is in image and video recognition, in particular in problems related to matching, representation and learning. He participated in a number of EU and UK projects in the area of image and video analysis. He publishes in computer vision, pattern recognition and machine learning forums. He has served in various roles at major international conferences co-chairing British Machine Vision Conference 2012, 2017 and IEEE International Conference on Advanced Video and Signal-Based Surveillance 2013. In 2014 he received Longuet-Higgins Prize awarded by the Technical Committee on Pattern Analysis and Machine Intelligence of the IEEE Computer Society.
Abstract
Matching is a process of finding correspondences (transforming, registering) between different sets of data (multiple images or videos, data from different sensors, times, depths, or viewpoints). It is used in robotics, computer vision, surveillance, medical imaging, remote sensing etc. I will present our recent works on image matching and localisation, how we were inspired by ideas proposed in the past, and how we implemented these ideas with the modern tools to advance the field. In past, significant attention was paid to the efficiency of new approaches for feature extraction, but this has been less targeted in deep learning techniques. I will discuss how the challenges have changed and how we address this issue. I will talk about the remaining problems in local feature extraction and their continuing relevance in computer vision applications, in particular large scale camera localisation, where classical methods and local features still remain competitive.
Keep on Learning
Tinne Tuytelaars
Electrical Engineering, KU Leuven
Belgium
http://www.esat.kuleuven.ac.be/psi/visics
Brief Bio
Tinne Tuytelaars is a full professor at KU Leuven, Belgium. Her research interests include computer vision and machine learning, with a special focus on representation learning, continual learning and multimodal learning (images and text). She has been one of the program chairs of ECCV14 and CVPR21 and one of the general chairs of CVPR16. She served as associate editor-in-chief of IEEE TPAMI for several years, and is on the editorial board of IJCV. In 2016, she received the Koenderink award at ECCV. In 2009 and 2021, she obtained a prestigious ERC starting/advanced researcher grant.
Abstract
Traditional machine learning assumes training and test data are sampled from the same distribution in an i.i.d. fashion.
However, in practice, this assumption is often violated. Therefore, in continual learning, alternative settings are studied,
where systems are to be updated with new knowledge over time, with minimal forgetting of what they have learnt in the past.
In this talk, I'll give an overview of some of our contributions to this field. In particular, I'll make a case for the use
of prototypical networks and the need for a more in-depth analysis of a network's behaviour when learning continually.
I'll finish with some open challenges and new continual learning settings beyond the standard task/class/domain incremental learning.
Image and Video Generation: A Deep Learning Approach
Nicu Sebe
University of Trento
Italy
Brief Bio
Nicu Sebe is a professor in the University of Trento, Italy, where he is leading the research in the areas of multimedia information retrieval and human-computer interaction in computer vision applications. He received his PhD from the University of Leiden, The Netherlands and has been involved in the past with the University of Amsterdam, The Netherlands and the University of Illinois at Urbana-Champaign, USA. He was involved in the organization of the major conferences and workshops addressing the computer vision and human-centered aspects of multimedia information retrieval, among which as a General Co-Chair of the IEEE Automatic Face and Gesture Recognition Conference, FG 2008, ACM International Conference on Multimedia Retrieval (ICMR) 2017 and ACM Multimedia 2013. He was a program chair of ACM Multimedia 2011 and 2007, ECCV 2016, ICCV 2017 and ICPR 2020. He is a general chair of ACM Multimedia 2022 and a program chair of ECCV 2024. Currently he is the ACM SIGMM vice chair, a fellow of IAPR and a Senior member of ACM and IEEE.
Abstract
Video generation consists of generating a video sequence so that an object in a source image is animated according to some external information (a conditioning label or the motion of a driving video). In this talk I will present some of our recent achievements adressing these specific aspects: 1) generating facial expressions, e.g., smiles that are different from each other (e.g., spontaneous, tense, etc.) using diversity as the driving force. 2) generating videos without using any annotation or prior information about the specific object to animate. Once trained on a set of videos depicting objects of the same category (e.g. faces, human bodies), our method can be applied to any object of this class. To achieve this, we decouple appearance and motion information using a self-supervised formulation. To support complex motions, we use a representation consisting of a set of learned keypoints along with their local affine transformations. A generator network models occlusions arising during target motions and combines the appearance extracted from the source image and the motion derived from the driving video. Our solutions score best on diverse benchmarks and on a variety of object categories.