Bálint Gyevnár
PhD CandidateCentre for Doctoral Training in Natural Language Processing;
University of Edinburgh.
I am interested in building trustworthy explainable AI for multi-agent systems, with applications to autonomous vehicles, and an additional focus on AI safety and trust.
My supervisors are:
Stefano Albrecht, Shay Cohen, and Chris Lucas.
Recent highlights
- New survey paper accepeted in IEEE T-ITS titled Explainable AI for Safe and Trustworthy Autonomous Driving: A Systematic Review;
- I gave two invited talks at the Charles University of Prague and the Czech Technological University on the fundamental problems of and potential solutions to classical XAI.
- Our paper was accepted at AAMAS-2024 on Causal Explanations for Sequential Decision-Making in Multi-Agent Systems;
- New workshop paper in EXTRAAMAS: People Attribute Purpose to Autonomous Vehicles When Explaining Their Behavior based on our new dataset of 14 scenarios and 1300+ annotated human-written explanations called HEADD: Human Explanations for Autonomous Driving Decisions;
- New publication at AAAI-24 about my vision for human-centric XAI.
I began researching how we might make better explanations for autonomous vehicles. I quickly realised along the way that the current state of explainable AI (XAI) is broken. Existing methods only work for some models with particular asumptions, they ignore OOD data, always assume that the AI model is correct, and so on…
Worst of all, most of XAI is unapproachable to most ordinary folks who have no expertise with AI systems. This inspired me to look into exploring better ways to create intelligible explanations that calibrate trust according to the abilities of the autonomous system.
I am interested in creating more trustworthy autonomous systems that are based on intelligible causal explanations and provide recourse through a conversational process.
As part of an interdisciplinary team, I work with people from fields such as reinforcement learning, cognitive science, natural language processing, and autonomous driving.
Tidbits about me
I often spend my free time learning languages. Currently, I speak five. In decreasing order of fluency, these are:
Hungarian (native), English (fluent), German (fluent), Japanese (conversational), and Russian (beginner).
I like playing volleyball and I am currently the vice president of the Edinburgh University Volleyball Club. I play the setter position. I also enjoy walking with people among the stark landscapes of the Scottish Highlands and taking some breathtaking photos while enduring harsh weather.
Occasionally, I sit down to practise the piano. At the moment, I am working through the second movement of Schubert’s piano sonata in B-flat major (D 960). Currently, I am reading “Gödel, Escher, Bach: An Eternal Golden Braid” by Douglas R. Hofstadter. A list of books I have read since I began keeping records is here.