A Brief Biography

Martin Ulbrich, Senior Expert CNECT.A.2. An economist by training, Martin has been working on digital issues for more than twenty years in the Commission from different angles. Most recently, he has joined the AI policy team in 2018, contributing to the drafting of the White Paper on AI and working on the AI Regulatory Framework (He is part of the team that wrote the White Paper). Before joining the AI unit in 2018 he had worked most recently on the impact of digitisation on the labour market, and before that on geoblocking and on the economics of networks.

Mr. Ulbrich has previously worked among others in the European Commission’s Joint Research Center, where he analysed ICT research across the EU,  as well as in its industrial policy and transport departments.

How to make Artificial Intelligence Ethical

He will present the Commissions policy approach in the follow-up to the White Paper on AI.

A Brief Biography

Full Professor of Econometrics at the University Ca’ Foscari of Venice. PhD in Applied Mathematics at the University Paris Dauphine.
Her main research field is Econometrics and its financial and economic applications. Her contribution are both methodological and applied. She contributed to the literature on simulation based methods and also on the Bayesian field. She devoted a lot of work to Markov switching models, pioneering their application in Finance and she continues to be a reference author for this type of models, in particular in the presence of latent components, which requires the use of simulation based techniques. She contributed to multivariate GARCH modelling, also integrating Markov switching components. In the last 10 years, she has devoted her research to analysing financial crises and systemic risk. Prof. Billio’s work on the field is nowadays very well recognized and she is often invited as an expert in conferences and round tables. The joint paper with Getmansky, Lo and Pelizzon (JFE 2012) is considered one of the main reference papers on the analysis of systemic risk and it draws the attention of academics, practitioners and regulators. In this paper, the idea of interconnectedness is introduced along with a relevant network structure. Starting from this initial work on networks, she built several lines of research to i) improve the network extraction; 2) use network topology for signalling and more recently 3) integrate the time dimension in order to develop dynamic network models. Moreover, in the last two years, she added the climate change dimension to her research agenda, being nowadays one of the main driver of financial stability.
Prof. Billio has published more than 100 technical papers in refereed journals, handbooks, and conference proceedings in the areas of econometrics and financial econometrics, with applications to risk measurement, volatility modelling, financial crisis and systemic risk. She is participating to many research projects financed by the European Commission, European Investment Bank, Eurostat and the Italian Ministry of Research (MIUR). She has been scientific coordinator of the SYRTO project, EU-FP7 project devoted to systemic risk measurement and she is local coordinator of three H2020-EE-CSA project on Energy Efficiency (EeMAP, EeDaPP and EeMIPP). She is also the coordinator of the EIBURS project “ESG-Credit.eu” dealing with the integration of ESG factors and climate change in Credit Analysis and Rating. The results of these and other research projects have appeared in peer-refereed journals including Journal of Econometrics, Journal of Financial Economics, Journal of Applied Econometrics, Journal of Financial Econometrics, Journal of Banking and Finance and European Journal of Operational Research.
Prof. Billio is actively involved in the organization of several scientific meetings and, in 2002 she co-established a new series of international workshops devoted to credit and financial risks (CREDIT), which has now reached the nineteenth edition (http://www.greta.it/credit/credit.htm). She is regularly on the program committees of the major international conferences and workshops of her fields and is currently member of the Board of Directors of the European Financial Management Association (EFMA) and member of the Scientific Committee of the Italian Association Financial Industry Risk Managers (AIFIRM).

Bayesian Dynamic Tensor Regression for multilayer financial networks

High dimension and multi-array data are becoming increasingly available in biology, physics, neuroimaging and economics. Examples include multi-layer economic and financial networks, multidimensional panels and brain images. These multidimensional data have a suitable representation as tensors and this calls for appropriate econometric tools which prevent data reshaping and are directly interpretable.
We propose a new dynamic linear model for tensor-valued response variables and covariates, called the tensor autoregressive model (ART), that encompasses some well known econometric models as special cases. Then, we derive on orthogonalized impulse response function, which allows for studying shock propagation within and between each dimension of the tensor-valued data.
We apply the ART model for analyzing the temporal evolution of multilayer networks of international trade and outstanding credit. The investigation is complemented by an impulse response analysis for studying the propagation of shocks across countries, over time and between layers. We found that, irrespective of its origin, any shock propagates between layers, but financial shocks are more persistent than those on international trade.

A Brief Biography

CEO and Founder of Thalesians Ltd. Previously served as Director and Head of global credit and core e-trading quants at Deutsche Bank, the teams that he helped set up with Jason Batt and Martin Zinkin. Having also worked at Morgan Stanley, Lehman Brothers, and Nomura, Paul pioneered electronic trading in credit with Rob Smith and William Osborn.

Paul has graduated from Christ Church, University of Oxford, with a distinction and Best Overall Performance prize. He has also graduated twice from Imperial College London.

Paul’s lectures at Imperial College London in machine learning for MSc students in mathematics and finance.

Paul has made contributions to mathematical logic, domain theory, and stochastic filtering theory, and, with Abbas Edalat, has published a prestigious LICS paper. Paul’s books are being published by Wiley, Springer, and World Scientific.

Toby Weston

Toby graduated with a Masters in Mathematics from Durham University, achieving distinction, in 2016. Since then he has spent two years working at JP Morgan Chase and is currently studying full time for an MSc in Mathematics and Finance at Imperial College London. His chosen thesis topic covers Reinforcement Learning for Optimal Execution.

Optimal Execution using Reinforcement Learning

Optimal execution involves finding a balance between the speed of execution and the minimisation of market impact. Prior work has shown that it has the potential to exploit idiosyncrasies of particular markets and beat basic benchmarks. Building upon this foundation we demonstrate the improvements that can be gained by applying recent developments in Reinforcement Learning to the trading environment.

A Brief Biography

Wolfgang Karl HÄRDLE attained his Dr. rer. nat. in Mathematics at Universität Heidelberg in 1982 and in 1988 his habilitation at Universität Bonn.  He is Ladislaus von Bortkiewicz Professor of Statistics at Humboldt-Universität zu Berlin and the director of the Sino German Graduate School (洪堡大学 + 厦门大学) IRTG1792 on “High dimensional non stationary time series analysis”.  He also serves as head of the joint BRC Blockchain Research Center (with U Zürich).  He is guest professor at WISE, Xiamen U, SMU, Singapore, NCTU, Hsinchu TW, Charles U, Prague CZ.

His research focuses on data sciences, dimension reduction and quantitative finance.  He has published over 30 books and more than 300 papers in top statistical, econometrics and finance journals. He is highly ranked and cited on Google Scholar, REPEC and SSRN. He has professional experience in financial engineering, smart (specific, measurable, achievable, relevant, timely) data analytics, machine learning and cryptocurrency markets. He has created a financial risk meter, FRM  hu.berlin/frm, a cryptocurrency index, CRIX thecrix.de. and organises regularly blockchainnights.com   His web page is: hu.berlin/wkh  

FRM the AI based Financial Risk Meter

(Andrija Mihoci, Michael Althof, Cathy Yi-Hsuan Chen, Wolfgang Karl Härdle)

A daily systemic risk measure is proposed accounting for links and mutual dependencies between financial institutions utilising tail event information. FRM (Financial Risk Meter) is based on Lasso quantile regression designed to capture tail event co-movements. The FRM focus lies on understanding active set data characteristics and the presentation of interdependencies in a network topology. Two FRM indices are presented, namely, FRM@Americas and FRM@Europe. The FRM indices detect systemic risk at selected areas and identifies risk factors. In practice, FRM is applied to the return time series of selected financial institutions and macroeconomic risk factors. Using FRM on a daily basis, we identify companies exhibiting extreme "co-stress", as well as "activators" of stress. With the SRM@EuroArea, we extend to the government bond asset class. FRM is a good predictor for recession probabilities, constituting the FRM-implied recession probabilities. Thereby, FRM indicates tail event behaviour in a network of financial risk factors.

A Brief Biography

Kay is managing partner at kinetic mind GmbH and has a background in quantitative finance, statistical analysis and software development of more than 15 years.

Prior to his current position, Kay worked as a Senior Quantitative Analyst for STEAG, the fifth largest German energy provider, for E.ON Energy Trading, one of Europe’s largest energy provider, and for Sal. Oppenheim, an Investment Bank in Frankfurt, Germany. Kay developed and implemented pricing and hedging functionalities for exotic derivatives on equities, precious metals and energy commodities. He was also responsible for the development of productively used statistical prediction models employing methods from time series analysis and statistical learning.

Kay graduated in Mathematics from the University of Frankfurt and holds a PhD in Mathematical Statistics from the University of Bochum.
As a Senior Research Associate at the University of Technology in Sydney, Australia, he worked on a project on hybrid commodity and interest rate modelling, as well as on exotic option pricing in stochastic volatility models. Kay follows the latest research in the areas of quantitative finance and statistical learning, and publishes regularly in peer reviewed journals.

Machine Learning for Data Generation and Hedging in Finance

In this talk some applications of Machine Learning methods for data generation and deep hedging in the context of trading and risk management are presented. The first part discusses the use of generative neural networks for creating realistic market data like forward curves and volatility surfaces. The benefit of this data augmentation is demonstrated in the second part by fitting a deep hedger and evaluating its performance under realistic market conditions.

A Brief Biography

Dr. Thrishantha Nanayakkara is an associate professor in Design Engineering and Robotics at Dyson School of Design Engineering (DSDE), Imperial College London , where he is also the Director of the Morph Lab. He has published more than 140 papers in flagship robotics conferences and journals including IEEE transactions on robotics, IEEE Robotics and Automation Letters, RSS, IROS, ICRA, and RoboSoft. He is in the executive committee of the UK RAS Strategic Task Group for Soft Robotics, and in the editorial board as an Associate Editor of flagship robotics publications such as IEEE Robotics and Automation Letters, RSS, ICRA, IROS, RoboSoft, Frontiers in Soft Robotics, and the Journal of Robotics and Mechatronics. He has worked at leading laboratories for robotics and neuromotor control, including the Laboratory for Computational Motor Control, Johns Hopkins University, MIT Computer Science and Artificial Intelligence Lab (CSAIL), and Harvard Neuromotor Control Lab. He is and has been PI on EPSRC and EU funded projects of more than £5 million that have pushed the boundaries of our understanding on how conditioning the body improves the efficacy of action and perception in human-human and human-robot interactions.

Shared computation between the brain and the body

A system is called an embedded system if it can take good enough actions in response to states within deadlines imposed by the environment. In that sense living beings and most robots are embedded systems. When states are uncertain, the task of state estimation within deadlines becomes non-trivial. Living beings often take a recursive approach to estimate such random variables. For instance, if someone is asked to estimate the weight of an object, they would bob it up and down several times before concluding an estimate. If we frame it as a Recursive Bayesian estimation process, the agent can significantly benefit from the ability to “morph” the likelihood function to sharpen the posterior distribution. In our studies we see that participants change the elbow stiffness and bobbing behavior depending on the weight of the object in the above scenario. We see similar phenomena in other estimation tasks too. In soft tissue palpation for instance, when a Physician is required to estimate the location of the edge of the liver of a patient using manual palpation, they would regulate the stiffness and configuration of the fingers to condition haptic perception during palpation. In this talk, I will show some recent results of this information morphing approach for efficient estimation of environmental states using a controllable stiffness body. I will show a soft robotic approach to test hypotheses we build based on human behavior.

A Brief Biography

Evelyn is a tech entrepreneur and Chief Customer Officer for the AI startup Mindfire. Her mission at Mindfire is empowering clients and partners with technology, and shaping the world with AI to the better. She has worked over a decade with fortune 500’s in various technology and innovation driven projects, and is active in advising tech startups globally. Her background is economics, she lives in Zurich.

On the quest for human level AI through virtual reality

What tech companies call AI is often just a race for better automation or brute force computing. 

To achieve human level Artificial Intelligence we need to be inspired by the renaissance geniuses, highly intelligent and transdisciplinary individuals. Mindfire‘s goal is to bring together some of the most creative individuals from different disciplines from around the globe in the world‘s largest virtual laboratory to create a new kind of collective superorganism or renaissance genius. 

The development of new, innovative intelligent systems that drive the creation of human level AI requires a radical change in AI research direction. 

A Brief Biography

Daan Kakebeeke is a Senior Manager at Bain & Company. In the past he helped pioneer one of the earliest and most successful data science products for Bain in Financial Services. More recently he is a founding member of their Industrial Analytics team – which includes industry experts, operations consultants and data scientists. His work centers on effectively scaling digital technology and enterprise AI applications for industrial firms globally in sectors such as Energy, Chemicals and Manufacturing. As part of the work, he regularly interacts with leading vendors in the ecosystem. He holds a Bsc. in Chemistry and MBA degree at the university of Berkeley, with a 1-year specialization in Computer Science.

Reinventing the Industrial base

Industrial firms have been slow to adopt enterprise AI applications versus many other sectors and for good reason. It is hard to make AI work at scale in an industrial setting, in a way that adds value to operators and the business. However for those firms that get it right, the positive impact on operational performance and sustainability can be a true step-change. This talk will provide a brief history of industry developments and then will focus on some of the most promising Industrial applications of AI now and in the future as the industry starts to reinvent itself.

A Brief Biography

Prof. Dr. Tim Weingärtner is a lecturer at the School of Information Technology at the Lucerne University of Applied Sciences and Arts (HSLU), Switzerland. He is working on blockchain technology and its applications in IoT, identity and the energy sector. Furthermore he organizes the International Blockchain Forum Rotkreuz (ibfr.ch). As a representative in the Smart-up Program, he supports the promotion of young start-ups from the HSLU. As a member of the project team, he played a major role in setting up the Central Switzerland Innovation Park in Rotkreuz. Under the thematic focus "Building Excellence", the park deals with the Digital Transformation in the construction industry.

Before joining the university, Tim worked in the Swiss financial industry for more than 15 years. He led several major IT projects in credit management and credit risk management. During this time, he also worked as a product manager for the leading provider of core banking solutions in Switzerland. Tim studied computer science and received his doctorate in medical robotics from the University of Karlsruhe, Germany.

Is DeReg the answer to DeFi? Impact of Blockchain on RegTech

Decentralized Finance (DeFi) refers to financial applications developed on the basis of blockchain systems and currently focuses mainly on monetary banking services, peer-to-peer lending and tokenization. The regulation of blockchain based financial applications faces various challenges such as country-specific rules in globalized systems, high uncertainty or lacking technological knowledge. In this ecosystem, which is becoming increasingly complex on all fronts, new approaches are essential. The presentation will stimulate the discussion in this area and highlight some possible approaches.

A Brief Biography

Dr. Andreas Theodorou is a postdoctoral researcher in the Responsible AI Group at Umeå University and the CEO and co-founder of VeRAI AB. His research interests include the development of software engineering methods of artificial intelligence, the verification and validation of ethical values in intelligent systems, the development means to provide transparency and explainability, and the study of the public’s perception of intelligent systems. In parallel to his research activities, Dr. Theodorou has been a contributing member of AI policy initiatives, e.g. IEEE SA’ P7001 series, ISO JTC1/42, UK’s AI APPG, EU’s AI Alliance, and others. He was part of the research team to evaluate the ethics guidelines for trustworthy AI suggested by the High-Level Expert Group on AI of European Commission. Dr. Theodorou has previously held research, visiting research, and teaching positions in the Georgia Institute of Technology (USA), University of Bath (UK), and University of Surrey (UK).

Contextualising AI Governance Guidelines

The last few years have seen a huge growth in the capabilities and applications of Artificial Intelligence (AI). Hardly a day goes by without news about technological advances and the societal impact of the use of AI. Not only are there large expectations of AI's potential to help to solve many current problems and to support the well-being of all, but also concerns are growing about the impact of AI on society and human wellbeing. Currently, many principles and guidelines have been proposed for “trustworthy” AI. They often rely on context-specific ethical socio-legal values — for example, fairness — but do not address the cultural variety between the different societies affected by AI system. Instead of more “one-size-fits-all” policies, in this talk I discuss socio-technical solutions to make such values explicit and, therefore, auditable.

A Brief Biography

Michael T. Stuart is a research fellow at the Centre for Philosophy of Science at the University of Geneva, where he is principal investigator on a research project concerning Artificial Imagination, and the role of AI in science. His PhD is from the Institute for the History and Philosophy of Science and Technology at the University of Toronto. He has since been a postdoctoral or visiting fellow at the Universities of Pittsburgh, Cambridge, Bielefeld, and the London School of Economics. He recently completed a fellowship at the Digital Society Initiative at the University of Zurich, and will soon begin a fellowship at the Carl Friedrich von Weizsäcker Center for Foundational Research at the University of Tübingen. Mike is pursuing sociological cum philosophical research on the degree to which machines can be said to “believe,” “be aware of,” “know,” “understand,” and “imagine.” Establishing which of these epistemic states can properly be predicated of artificial agents is important for debates concerning our ability to praise and blame these agents for their “actions,” as well as for developing legal and moral frameworks for the attribution of (at least partial) responsibility when things go wrong (or right).

Markus Kneer (MA: Oxford University, PhD: Ecole Normale Supérieure, Paris) is the Principle Investigator of the Guilty Minds Lab (Centre for Ethics, University of Zurich) and a fellow at the Digital Society Initiative (UZH). After research positions at Pittsburgh University and Columbia University he joined the University of Zurich, where he works on theory of mind, ethics, and AI broadly conceived. Currently, he is particularly interested in biases in criminal trials on the one hand, and the psychological fundamentals of human-robot interaction on the other.   

Artificial Responsibility

Tomas Hauer (2020) identifies two strands of work in the ethics of AI: one which creates and applies ethical rules for AIs, and another which asks whether AIs could behave ethically, in principle. To this second question, Kestutis Mosakas (forthcoming) has recently argued that artificial agents cannot behave ethically because AIs would need to be conscious to be ethical persons. Hauer claims that we should give up on this second question because the dominant methodology is a priori arguments and thought experiments about deep philosophical concepts (like consciousness and intentionality) that will not be resolved in time to assist in answering this question. We disagree, and see value in the efforts of some to make more piecemeal progress by considering less than fully autonomous entities as potentially partially responsible. We agree with Hauer that the way forward should not be through a priori arguments and unrealistic thought experiments, so we adopt an experimental method, testing the analogies drawn between AIs, animals, and corporations (e.g., by Laukyte 2020) with respect to moral responsibility and blameworthiness, and especially the capacity to have a "mens rea."  

Hauer, Tomas. 2020. ‘Machine Ethics, Allostery and Philosophical Anti-Dualism: Will AI Ever Make Ethically Autonomous Decisions?’ Society, July. https://doi.org/10.1007/s12115-020-00506-2.
Laukyte, Migle. 2020. ‘The Intelligent Machine: A New Metaphor through Which to Understand Both Corporations and AI’. AI & Society, July. https://doi.org/10.1007/s00146-020-01018-7.

Mosakas, Kestutis. forthcoming. ‘On the Moral Status of Social Robots: Considering the Consciousness Criterion’. AI and Society, 1–15. https://doi.org/10.1007/s00146-020-01002-1.