How Theoretical Analyses Can Impact Practical Applications of Evolutionary Computation
Pietro S. Oliveto, University of Sheffield, United Kingdom
Swarm Intelligence in Distributed Systems Use-cases
Vesna Sesum-Cavic, Institute of Information Systems Engineering, TU Vienna, Austria
From Machine Learning to Explainable AI and Beyond
Andreas Holzinger, Human-Centered AI Lab, Institute for Medical Informatics and Statistics, Medical University Graz, Austria
Type-2 Fuzzy Systems for Human Decision Making
Jonathan Garibaldi, School of Computer Science, University of Nottingham, United Kingdom
How Theoretical Analyses Can Impact Practical Applications of Evolutionary Computation
Pietro S. Oliveto
University of Sheffield
United Kingdom
Brief Bio
Pietro S. Oliveto is a Senior Lecturer and EPSRC Early Career Fellow at the Department of Computer Sci-ence, University of Sheffield where he leads the ’Rigorous Runtime Analysis of Bio-inspired Computing’ project team, ‘Rigorous Research’ in short.
He received the Laurea degree in computer science from the University of Catania, Italy in 2005 and the PhD degree in computational complexity of evolutionary algorithms from the University of Birmingham, UK in 2009. He has been an EPSRC PhD Plus Fellow (2009-2010) and an EPSRC Postdoctoral Fellow in Theo-retical Computer Science (2010-2013) at Birmingham and a Vice-Chancellor’s Fellow (2013-2016) at Shef-field.
Dr. Oliveto is Chair of the IEEE Computational Intelligence Society (CIS) Technical Committee for Evolu-tionary Computation (ECTC). He is Associate Editor of IEEE Transactions on Evolutionary Computation (IEEE TEVC) and Editorial Board member of the Algorithms Journal. He has edited special issues of the IEEE TEVC, Evolutionary Computation (ECJ) and Theoretical Computer Science (TCS) journals. He has Chaired the annual Symposium on Foundations of Computational Intelligence (IEEE FOCI) (2014-2020) and is a Steering Committee member of the workshop series on Theory of Randomised Search Heuristics (ThRaSH). He leads the benchmarking Working Group (WG3) of the COST Action CA15140: ‘Improving Applicability of Nature-Inspired Optimisation by Joining Theory and Practice’. He is a member of the EPSRC Peer Review College and of the Carnegie Trust of Research Assessors for the Universities of Scotland.
Abstract
Rigorous computational complexity analyses of evolutionary algorithms have been performed since the nineties. The first results were inevitably related to simplified algorithms for benchmark problems with significant structures. Nowadays analyses are possible for classical combinatorial optimisation problems with real world applications and for the standard bio-inspired optimisation algorithms that are used in practice. Such analyses shed light on which classes of algorithms should be preferred for different problems and on how to set the algorithmic parameters. In this talk I will provide examples of the insights which can be gained from these theoretical analyses and how they can ultimately lead to better algorithms, parameter settings and results in practice.
Swarm Intelligence in Distributed Systems Use-cases
Vesna Sesum-Cavic
Institute of Information Systems Engineering, TU Vienna
Austria
Brief Bio
Vesna Šešum-Cavic is a Senior Scientist and University Lecturer in Computational Intelligence, Institute of Information Systems Engineering, Compilers and Languages Group, Vienna University of Technology, Austria. She is a graduated mathematician (Dipl.Math) and Magistar of Computer Science (Mag.) from University of Belgrade. She received a doctoral degree in Computer Science from Vienna University of Technology (Dr. techn.). Her research interests cover swarm intelligence, network optimization, p2p systems, theory and design of algorithms, combinatorial optimization, complex systems, self-organization, multi-agent systems. She was a conference chair/program committee member of international conferences. Vesna is the current Chair of IEEE Women in Computational Intelligence, and a member of the IEEE Women in Engineering Committee, IEEE CIS Webinars sub-committee and the IEEE CIS Member Activities Committee.
Abstract
The growing complexity of nowadays distributed systems becomes a critical issue. Distributed software systems are forced to integrate other software systems and components that are often not reliable, exhibit bad performance, and are sometimes unavailable. Such software is typically characterized by a huge problem size concerning number of computers, clients, requests and size of queries, autonomy and heterogeneity of participating organizations, and dynamic changes of the environment. To cope with unforeseen dynamics in the environment and vast number of unpredictable dependencies on participating components, there is a demand on self-organizing approaches.
Swarm intelligence possesses distributive and autonomous properties, and represents a self-organizing biological system. Therefore, swarm-inspired algorithms play an important role in the design of self-organizing software for distributed systems. Depending on the problem area, an application of swarm-inspired algorithms enables different kinds of self-organization.
In this keynote lecture, an overview of the significance and power of swarm intelligence in coping with some typical distributed systems problems (e.g, information retrieval, load balancing, load clustering, distributed routing) will be presented. It also includes some examples of their implementation and adaptation to these use-cases.
From Machine Learning to Explainable AI and Beyond
Andreas Holzinger
Human-Centered AI Lab, Institute for Medical Informatics and Statistics, Medical University Graz
Austria
https://www.aholzinger.at/
Brief Bio
Andreas Holzinger is lead of the Holzinger Group (Human-Centered AI) at the Medical University Graz and Visiting Professor for explainable AI at the Alberta Machine Intelligence Institute in Edmonton, Canada. Since 2016 he is Visiting Professor for Machine learning in health informatics at Vienna University of Technology. Andreas was Visiting Professor for Machine Learning and Knowledge Extraction in Verona, RWTH Aachen, University College London and Middlesex University London. He serves as consultant for the Canadian, US, UK, Swiss, French, Italian and Dutch governments, for the German Excellence Initiative, and as national expert in the European Commission. Andreas obtained a Ph.D. in Cognitive Science from Graz University in 1998 and a second Ph.D. (Habilitation) in Computer Science from TU Graz in 2003. Andreas Holzinger works on Human-Centered AI (HCAI), motivated by efforts to improve human health. Andreas pioneered in interactive machine learning with the human-in-the-loop. For his achievements, he was elected as a member of Academia Europea in 2019. Andreas is paving the way towards multimodal causability, promoting robust interpretable machine learning, and advocating for a synergistic approach to put the human-in-control of AI and align AI with human values, privacy, security, and safety.
Abstract
Explainable AI is rapidly becoming extremely important. Whilst classic rule-based approaches of early AI have been comprehensible "glass-box" approaches at least in narrow domains, their weakness was in dealing with uncertainties of the real world. The introduction of statistical/probabilistic learning methods has made AI increasingly successful. Meanwhile deep learning approaches even exceed human performance in particular tasks. However, such approaches are becoming increasingly opaque, and even if we understand the underlying mathematical principles of such models, they still lack explicit declarative knowledge. For example, words are mapped to high-dimensional vectors, making them unintelligible to humans. What we need in the future are context-adaptive procedures, i.e. systems that construct contextual explanatory models for classes of real-world phenomena. This talk presents some recent approaches on how to understand why a machine decision has been reached, making results re-traceable, explainable and comprehensible on demand and paving the way for an expert in the loop to augment human intelligence with artificial intelligence.
Type-2 Fuzzy Systems for Human Decision Making
Jonathan Garibaldi
School of Computer Science, University of Nottingham
United Kingdom
http://www.cs.nott.ac.uk/~jmg/home.shtml
Brief Bio
Prof Jon Garibaldi's main research interest is in the modelling of human decision making, primarily in the context of medical applications. His work to date has concentrated on utilising fuzzy logic to model the imprecision and uncertainty inherent in medical knowledge representation and decision making. This has been applied in areas such as classification of breast cancer, identification of Alzheimer's disease, and the assessment of immediate neonatal outcome. A particular interest is in the transfer of medical intelligent systems into clinical use and this has led to the study of methods of evaluating intelligent systems and mechanisms for their implementation. Prof Garibaldi also has an interest in generic machine learning, such as clustering, classification and optimisation, particularly when applied to the optimisation of decision making models, and in the study of adaptive and time-varying behaviour.
Abstract
Type-2 fuzzy sets and systems, including both interval and general type-2 sets, are now firmly established as tools for the fuzzy researcher that may be deployed on a wide range of applications and in a wide set of contexts. However, in many situations the output of type-2 systems are type-reduced and then defuzzified to an interval centroid, which are then often even simply averaged to obtain a single crisp output. Many successful applications of type-2 have been in control contexts, often focussing on reducing the RMSE. This is not taking full advantage of the extra modelling capabilities inherent in type-2 fuzzy sets. In this talk, I will present some of the current research being carried out at Nottingham into type-2 for modelling human reasoning. I will cover approaches and methodologies which make more use of type-2 capabilities, illustrating these with reference to practical applications such as modelling expert variability, classification of breast cancer tumours, cyber-security contexts, and other decision support problems.