AXA Chair in Adversarial Risk Analysis

The AXA-ICMAT Chair in Adversarial Risk Analysis ran from September 2014 to August 2023 and resided at the Institute of Mathematical Sciences of the Spanish National Research Council. The holder was Prof. David Rios Insua, currently Research Professor of Risk Analysis and Data Science at ICMAT-CSIC. Its objectives were to develop the foundations, methods, computations, and applications in Adversarial Risk Analyisis and improve the transfer capabilities of ICMAT. Adversarial risk analysis is a recent methodological paradigm that expands probabilistic risk analysis to handle risks emerging from intelligent adversaries as in homeland security, cybersecurity, or business competition.

An overview of the topics covered by the chair include:

The Chair has been successful  in many methodological advances. Some of the key ones are: i) A systematic analysis of ARA methods, including a comparison with core standard game theoretic concepts and behavioral experiments, showcasing its relevance; ii) a fast and scalable methodology to compute optimal ARA solutions (and Nasheq) in sequential games; iii) a radically new approach to adversarial machine learning based on ARA, applied to classification and reinforcement learning  approaches; iv) novel Bayesian computation methods for very large scale models, including deep learning models;  v) a new methodology for affective decision making; the initial steps of a novel adversarial statistical decision theory.

The Chair has also been successful in numerous applications including: i) A novel methodology for cybersecurity risk management overcoming defects of current standards and facilitating optimal resource allocation, including cyberinsurance and a protoype tool; ii) a novel method for aviation safety risk management fully tested for the Spanish government in the RIMAS system with enormous safety savings; iii) models and software to predict risks in relation to cardiovascular diseases and colorectal cancer; iv) adversarial risk analysis models to support automated driving systems; v) novel decision analytic models for drug discovery. Many of them could lead to industrial products.

The Chair has led to 93 papers in major jorunals (such as JASA, Stat. Science, EJOR, Risk Analysis,…), 10 books, 28 additional  projects (entailing 3.6M€ additional funding). It has supported 39 new researchers (BSc 4, MSc 6,  PhD 14, Postdocs 15), 38 visitors and 1 research manager. Important prizes (like DeGroot from ISBA, Edelman from INFORMS, or from SEIO) were awarded to the Chair, who also undertook important divulgation activities including the It’s a risky life YouTube series, the  ICMAT Machine Learning YouTube series and 8 articles at El País.