Early Stage Researcher at UNIABDN

Adarsa is a machine learning practitioner experienced in open-source development, primarily in applied Natural Language Processing. She has worked on system building at scale with more than seven years of industry experience. She currently explores the topic of Explainable Bayesian Networks at University of Aberdeen.

UNIABDN - University of Aberdeen (United Kingdom) Department of Computing Science

Department of Computing Science
University of Aberdeen (UNIABDN)
Kings College, Aberdeen
AB24 3UE, United Kingdom

Explaining Probabilistic Reasoning

PhD research topic


To develop techniques for explaining Bayesian and probabilistic reasoning, especially Bayesian networks, to different types of users. Bayesian reasoning is transparent and sound, but very different from human reasoning, especially amongst people who are not domain experts. We will empirically explore the difficulties that different types of users have in understanding Bayesian networks; create an algorithm for generating appropriate explanations for different types of users; and empirically evaluate the effectiveness of this algorithm.

Expected Results:

– In-depth understanding of the state of the art in explaining Bayesian reasoning, especially in Bayesian networks.

– Analysis of the difficulties different types of people have in understanding Bayesian reasoning.

– Algorithm for generating appropriate explanations for different users of Bayesian reasoning.

– Implementation of the algorithms, and integration into a standard tool for examining Bayesian networks.

– Evaluation of the effectiveness of the algorithm in increasing user understanding, trust, and decision making.

– Collaboration with and supporting other NL4XAI- ESRs working on related topics, including NL4XAI- ESR2, NL4XAI-ESR4, and NL4XAI-ESR10