MOD 2016 Keynote Speakers
“From Pattern Recognition to Artificial Intelligence”,
Nello Cristianini, University of Bristol, UK,
Modern Artificial Intelligence is powered by statistical methods and fueled by large amounts of data. This reliance on machine learning enabled us to bypass the need to fully understand a phenomenon before replicating it into a computer, paving the way to much progress in fields as diverse as machine translation, computer vision, speech recognition. This approach is affecting also other disciplines: we call this the big-data revolution. For this reason, data has been called the new oil: a new natural resource, that businesses and scientists alike can leverage. Along with great benefits, this approach comes also with many risks, some linked to the changes it brings to scientific method, others due to the fact that very often the valuable data being used is our own personal data. The current trend towards collection, storage, analysis and exploitation of large amounts of personal data needs to be understood, its implications need to be assessed, particularly those that will affect our autonomy, our rights, public opinion, and other fundamental aspects of our life.
Bio sketch:
Nello Cristianini is a Professor of Artificial Intelligence at the University of Bristol, His current research covers the large scale analysis of media content (news and social media), using various AI methods, and the implications of Big Data. Cristianini is the co-author of two widely known books in machine learning, “An Introduction to Support Vector Machines” and “Kernel Methods for Pattern Analysis” and of a book in bioinformatics “Introduction to Computational Genomics”. He is also a recipient of the Royal Society Wolfson Research Merit Award and a current holder of a European Research Council Advanced Grant.
“High-dimensional Vector Autoregressive Processes: Modeling, Estimation and Applications”,
George Michailidis, University of Florida, USA
Vector Autoregressive processes represent a popular class of time series models that aim to capture temporal interconnections between a number of time series. They have been widely used in economics and finance and more recently in biomedical applications. In this talk, we discuss modeling and estimation issues in the high-dimensional setting under different constraints on the transition matrices –sparsity, low rankness, etc. We discuss optimization and inference issues and illustrate the results with applications to financial stability and biological regulation.
Bio sketch: TBA
“Deep Symbolic Learning”,
Stephen Muggleton, Imperial College London, UK,
The term Deep Learning is generally associated with estimation techniques for multi-layered neural networks. However, recent developments in symbolic learning techniques have shown increased accuracy and efficiency gains can be achieved by use of multi-layered learning within logical representations. In particular, this presentation will describe the new area of Meta-Interpretive Learning (MIL), which supports automatic introduction of sub-definitions when learning predicate definitions, allowing decomposition into a hierarchy of reuseable parts. In applications involving the learning of string transformations MIL has been used to construct multi-layered networks involving a variety of
automatically invented relations. MIL is based on an adapted version of a Prolog meta-interpreter. Normally such a meta-interpreter derives a proof by repeatedly fetching first-order Prolog clauses whose heads unify with a given goal. By contrast, a meta-interpretive learner additionally fetches higher-order meta-rules whose heads unify with the goal, and saves the resulting meta-substitutions to form a program. This talk will overview theoretical and implementational advances in this new area including the ability to learn Turing computabale functions within a constrained subset of logic programs, the use of probabilistic representations within Bayesian meta-interpretive and techniques for minimising the number of meta-rules employed. The talk will also summarise applications of MIL including the learning of regular and context-free grammars, learning from visual representions with repeated patterns, learning string transformations for spreadsheet applications, learning and optimising recursive robot strategies and learning tactics for proving correctness of programs. The talk will conclude with a discussion
of the potential for integrating symolic and sub-symbolic learning within the context of a Meta-Interpretive Learning framework.
Bibliography
A. Cropper and S.H. Muggleton. Learning higher-order logic programs through abstraction and invention. In Proceedings of the 25th International Joint Conference Artificial Intelligence (IJCAI 2016), pages 1418-1424. IJCAI, 2016
A. Cropper and S.H. Muggleton. Learning efficient logical robot strategies involving composable objects. In Proceedings of the 24th International Joint Conference Artificial Intelligence (IJCAI 2015), pages 3423-3429. IJCAI, 2015.
W-Z Dai, S.H. Muggleton, and Z-H Zhou. Logical vision: Meta-interpretive learning for simple geometrical concepts. In Short Paper Proceedings of the 25th International Conference on Inductive Logic Programming. National Institute of Informatics, Tokyo, 2015.
S.H. Muggleton, D. Lin, and A. Tamaddoni-Nezhad. Meta-interpretive learning of higher-order dyadic datalog: Predicate invention revisited. Machine Learning, 100(1):49-73, 2015.
D. Lin, E. Dechter, K. Ellis, J.B. Tenenbaum, and S.H. Muggleton. Bias reformulation for one-shot function induction. In Proceedings of the 23rd European Conference on Artificial Intelligence (ECAI 2014), pages 525-530, Amsterdam, 2014. IOS Press.
Bio sketch: TBA
“A new Information Theory perspective on Network Robustness”,
Panos Pardalos, University of Florida, USA
A crucial challenge in network theory is the study of the robustness of a network when facing a sequence of failures. We propose a novel methodology to measure the robustness of a network to component failures or targeted attacks based on Information Theory, that considers measurements of the structural changes caused by failures of the network’s components providing a dynamical information about the topological damage. The methodology is comprehensive enough to be used with different probability distributions
and provides a dynamic profile that shows the response of the network’s topology to each event, quantifying the vulnerability of these intermediate topologies.
Bio sketch: TBA