10.00 – 10.30

Arrival, Refreshments

10.30 – 10.45

Welcome by Professor Ioannis Kosmidis, Turing University Lead at the University of Warwick

10.45 – 11.00

Professor Mark Girolami, Chief Scientist of The Alan Turing institute

Talk Title: Science and Innovation strategy to date

Turing Profile Page

11.00 - 11.15

Professor Jinglai Li, School of Mathematics, University of Birmingham

Talk Title: Entropy Estimation via Normalizing Flow

Abstract: 

Entropy estimation is an important problem in information theory and statistical science. Many popular entropy estimators suffer from fast growing estimation bias with respect to dimensionality, rendering them unsuitable for high dimensional problems. We propose a transform-based method for high dimensional entropy estimation, which consists of the following two main ingredients.

Firstly, by modifying the k-NN based entropy estimator, we propose a new estimator which enjoys small estimation bias for samples that are close to a uniform distribution.

Secondly, we will design a normalizing flow based mapping that pushes samples toward a uniform distribution, and the relation between the entropy of the original samples and the transformed ones is also derived. As a result the entropy of a given set of samples is estimated by first transforming them toward a uniform distribution and then applying the proposed estimator to the transformed samples. Numerical experiments demonstrate the effectiveness of the method for high dimensional entropy estimation problems.

University Profile Page

Turing Profile Page

11.15 – 11.30

Break, Refreshments

11.30 – 11.45

Dr Max Little, School of Computer Science, University of Birmingham

Talk Title: Causal Bootstrapping: Plug-in Interventional Machine Learning

Abstract:

To draw scientifically meaningful conclusions and reliable inferences about quantitative phenomena, machine learning must take cause and effect into consideration (either implicitly or explicitly). This is particularly challenging when the relevant measurements are not obtained from controlled experimental (interventional) settings, so that cause and effect can be obscured by spurious, indirect influences. Modern predictive techniques from machine learning are capable of capturing high-dimensional, complex nonlinear relationships between variables while relying on few parametric or probabilistic modelling assumptions. However, since these techniques are associational, applied to observational data they are prone to picking up spurious influences from non-experimental (observational) data, making their predictions unreliable.

Techniques from causal inference, such as probabilistic causal diagrams and do-calculus, provide powerful (nonparametric) tools for drawing causal inferences from such observational data. However, these techniques are often incompatible with modern, nonparametric machine learning algorithms since they typically require explicit probabilistic models.

In this talk I'll describe causal bootstrapping, a new set of techniques we have developed for augmenting classical nonparametric bootstrap resampling with information about the causal relationship between variables. This makes it possible to resample observational data such that, if it is possible to identify an interventional relationship from that data, new data representing that relationship can be simulated from the original observational data. In this way, we can use modern statistical machine learning and signal processing algorithms unaltered to make statistically powerful, yet causally-robust, inferences.

University Profile Page

Turing Profile Page

11.45 - 12.00

Dr Manuel Lopez-Ibanez, Manchester Business School, University of Manchester

Talk Title: Auto-Optimisation: Automatising the Design of Algorithms for Solving Complex Optimisation Problems

Abstract:

Optimisation problems arise in many domains, including management, logistics, manufacturing, engineering, biology and chemistry.

There are numerous methods for solving every type of optimisation problem. Such methods are often the result of combining specific algorithmic components and tuning their hyper-parameters. Therefore, when facing a problem, the main challenge is often to choose the right method and its components. Although the concepts of hyper-parameter optimisation and automatic algorithm selection were initially developed in the optimisation field, they have become far more popular and successful in the machine learning community, thus giving birth to the emerging field of AutoML.

University Profile Page

Turing Profile Page

12.00 – 12.15

Dr Paolo Turrini, Department of Computer Science, University of Warwick

Talk Title: Network Dynamics and the Emergence of Pro-social Behaviour

Abstract:

The emergence of pro-social (elsewhere called other-regarding or cooperative) behaviour among self-interested agents has been a key concern of the multi-agent systems community for decades.

With the increased importance of network-mediated interaction, researchers have shifted the attention on the impact of social networks and their dynamics in promoting or hindering cooperation, drawing various context-dependent conclusions. For example, some lines of research suggest the existence of a threshold effect in the ratio of timescales of network evolution, after which co-operation will emerge, whereas other lines dispute this, suggesting instead a Goldilocks zone.

In this talk I will use an evolutionary game theory framework to understand coevolutionary processes from a bottom-up perspective - in particular the emergence of a cooperator-core and defector-periphery - clarifying the impact of partner selection and imitation strategies in promoting cooperative behaviour, without assuming underlying communication or reputation mechanisms.

University Profile Page

Turing Profile Page

12.15 – 12.30

Professor Mehrdad Dianati, WMG, University of Warwick

Talk Title: Achieving Higher Levels of Vehicle Autonomy: Challenges and Potentials

Abstract:

Connected Autonomous Systems promise enormous opportunities in various applications to address some of the grand challenges of the 21st century. The term refers to groups of autonomous systems working together and with humans coherently and cohesively to extend humans’ physical and intellectual capabilities.

Such systems can unleash the power of big data and AI and processing and storage of the cloud (Edge/Core) and connectivity capabilities of 5G/6G communication systems to help optimise offline and/or in various real-time various aspects of the businesses and human activities. However, these systems must demonstrate a higher level of autonomy to realise their full potential. At the same time, they shall be designed, verified/validated to be trustworthy to be deployed safely and effectively in mass in real-world applications. Trust is a multi-facet and cross-disciplinary challenge. Trust is a function of competence and integrity of the systems and individuals.

An essential element of competence is resilience, defined as the capability of a system to avoid, withstand, adapt to, and recover from system failures. This talk will focus on some promising ideas that distributed intelligence can offer by harvesting the benefits of big data, core/edge cloud computing, and AI, expected to be enabled by the emerging services of the future 5G/6G communication systems in addressing the challenge of resilience in connected/cooperative autonomous systems.

University Profile Page

Turing Profile Page

12.30 – 14.00

Networking Lunch