Dr. Eoin Delaney
Assistant Professor, Computer Science
Biography
Dr. Eoin Delaney is an Assistant Professor in Trinity College Dublin and is an expert in responsible machine learning, encompassing topics such as algorithmic fairness, explainability and sustainability. He consistently publishes in top machine learning conferences (NeurIPS, AAAI, IJCAI) and journals. He has won two best paper awards and a national award for the Best Application of AI in a Student Project for work on interpretable machine learning with applications in sustainable smart agriculture. He has extensive experience with deep learning and machine learning frameworks and in designing large scale user studies for evaluating explanations of AI systems on Prolific. Previously Eoin completed his postdoc at the University of Oxford where he maintains active collaborations. He has designed open source python toolkits such as OxonFair and leads a research team focused on responsible and reliable machine learning systems. Eoin is especially interested in the deployment and evaluation of machine learning systems in real world scenarios.
Publications and Further Research Outputs
Peer-Reviewed Publications
Kai Rawal, Zihao Fu, Eoin Delaney, Chris Russell, Evaluating Model Explanations without Ground Truth, ACM Conference on Fairness, Accountability, and Transparency, Athens, Greece, 2025
Harry Mayne, Ryan Othniel Kearns, Yushi Yang, Andrew M Bean, Eoin Delaney, Chris Russell, Adam Mahdi, LLMs Don't Know Their Own Decision Boundaries: The Unreliability of Self-Generated Counterfactual Explanations, Empirical Methods in Natural Language Processing (EMNLP), Suzhou, China, 4/11/2025, 2025
Fu, Zihao and Brown, Ryan and Shao, Shun and Rawal, Kai and Delaney, Eoin and Russell, Chris, FairImagen: Post-Processing for Bias Mitigation in Text-to-Image Models, Advances in Neural Information Processing Systems (NeurIPS), 2025
Warren, Greta and Delaney, Eoin and Gueret Christophe, and Keane M.T, Explaining multiple instances counterfactually: User tests of group-counterfactuals for xai, International Conference on Case-Based Reasoning, International Conference on Case-Based Reasoning ICCBR, 2024, pp206--222
Delaney, Eoin and Fu, Zihao and Wachter, Sandra and Mittelstadt, Brent and Russell, Chris, OxonFair: A Flexible Toolkit for Algorithmic Fairness, Advances in Neural Information Processing Systems (NeurIPS), 2024
Kenny, Eoin and Delaney, Eoin and Keane, Mark, Advancing Post Hoc Case Based Explanation with Feature Highlighting, International Joint Conference on Artificial Intelligence (IJCAI-21), International Joint Conference on Artificial Intelligence (IJCAI-21), 2023
User tests & techniques for the post-hoc explanation of deep learning in, Explainable Deep Learning AI, Elsevier, 2023, pp263--291 , [Delaney, Eoin and Kenny, Eoin M and Greene, Derek and Keane, Mark T]
Delaney, Eoin and Pakrashi, Arjun and Greene, Derek and Keane, Mark T, Counterfactual explanations for misclassified images: How human and machine explanations differ, Artificial Intelligence, 324, 2023, p103995
Delaney, Eoin and Greene, Derek and Shalloo, Laurence and Lynch, Michael and Keane, Mark T, Forecasting for sustainable dairy produce: enhanced long-term, milk-supply forecasting using k-NN for data augmentation, with prefactual explanations for XAI, International Conference on Case-Based Reasoning, International Conference on Case-Based Reasoning, 2022, pp365--379
Kenny, Eoin M and Delaney, Eoin D and Greene, Derek and Keane, Mark T, Post-hoc explanation options for xai in deep learning: The insight centre for data analytics perspective, Explainable Deep Learning AI, Explainable Deep Learning AI, 2021, pp20--34
Delaney, Eoin and Greene, Derek and Keane, Mark T, Uncertainty estimation and out-of-distribution detection for counterfactual explanations: Pitfalls and solutions, ICML-21 Workshop on Algorithmic Recourse, 2021
Keane, Mark T and Kenny, Eoin M and Delaney, Eoin and Smyth, Barry, If only we had better counterfactual explanations: Five key deficits to rectify in the evaluation of counterfactual xai techniques, 30th International Joint Conference on Artificial Intelligence (IJCAI-21), IJCAI, 2021
Wallace, Duncan and Delaney, Eoin and Keane, Mark T and Greene, Derek, Nearest Neighbour-Based Data Augmentation for Time Series Forecasting., AICS, Explainable Deep Learning AI, 2021, pp60--71
Delaney, Eoin and Greene, Derek and Keane, Mark T, Instance-based counterfactual explanations for time series classification, International conference on case-based reasoning, 2021, pp32--47
Non-Peer-Reviewed Publications
Goethals, Sofie and Delaney, Eoin and Mittelstadt, Brent and Russell, Chris, Resource-constrained fairness, arXiv preprint arXiv:2406.01290, 2024
Kaivalya Rawal, Eoin Delaney, Zihao Fu, Sandra Wachter, Chris Russell, Evaluating the Ability of Explanations to Disambiguate Models in a Rashomon Set, AAAI 2026, 1st Workshop on Navigating Model Uncertainty and the Rashomon Effect: From Theory and Tools to Applications and Impact, AAAI 2026
Research Expertise
Description
My research vision is to build and evaluate responsible, interpretable and human centred systems that drive ethical and sustainable change in deployment. To realize this vision, my long-term goal is to establish a diverse, world-leading research lab that publishes in top venues, actively collaborates with industry and global partners, and engages with the public to champion equality, diversity, and inclusion. My research plan aligns with the E3 Balanced Solutions for a Better World framework, specifically under the themes of sustainability and ethical data use, two related areas that underpin my vision and research expertise in responsible machine learning. Through interdisciplinary research during my research career, I have taken steps towards developing these systems by designing trustworthy algorithms and open-source toolkits, applying them to critical applications in sustainability and ethical data use, and evaluating their impact on end users. Explainable Predictions in AI with Applications in Sustainability. Ongoing and previous collaborations with industry (Accenture Labs, Glanbia, Google Deepmind, The Central Bank of Ireland) and domain experts in applied domains such as Finance and Smart Agriculture (Teagasc) have enabled me to lead interdisciplinary projects that tackle real-world problems with robust and transparent solutions in responsible machine learning. One example of this was my work on eXplainable AI (XAI) for sustainable smart agriculture. My applied work on providing long-term and accurate milk supply forecasts to dairy farmers demonstrated that providing explanations could boost their on-farm performance in terms of increased yield and sustainability. These explanations could leverage high-performing farms as gold-standard prototypes to provide a basis for informing farmers on best practices. This work won a best paper award, and I received a national award at the AI Ireland Awards for the Best-Application of AI in a Student Project. In future work, I would like my team to explore the stability and robustness of explanations over time. In real-world scenarios, there is no guarantee that new data will come from the same distribution as previous observations. This can be detrimental for the reliability of explanations and is especially relevant if predictions are influenced by shifts such as climate change. Moreover, as explanations are ultimately for end-users, it is crucial to consider how they can impact trust in automated decision-making. Ethical Data Use with Applications in Algorithmic Fairness Responsible machine learning encompasses a broader spectrum of principles beyond explainability, including reliability, fairness, accountability, and ethical considerations. My research goal is not only to build and evaluate explainable systems but also to ensure that they are reliable, fair, and ethical. In my recent NeurIPS paper (Core A*), I designed an open-source software toolkit to measure and enforce fairness with a focus on high-capacity models (deep neural networks) in classification scenarios. Our subsequent work on text-image generative models has been accepted to NeurIPS 2025.Projects
- Title
- Central Bank PhD Programme in Artificial Intelligence & Data Science (Explanation of AI)
- Funding Agency
- Central Bank of Ireland and Research Ireland
Recognition
Representations
Programme committee member for Neural Information Processing Systems (NeurIPS)
Programme committee member for ACM Conference on AI Ethics and Society Conference (AIES)
Reviewer for Artificial Intelligence Journal (AIJ) - Top AI Journal
Programme committee member for ACM Conference on Fairness, Accountability, and Transparency (FAccT)
Programme committee member for ACM Knowledge Discovery in Databases Conference (ACM SIGKDD)
Awards and Honours
Best Application of AI in a Student Project - Irish National Award Winner
Best Paper - ICCBR 2022
Best Student Paper - ICCBR 2021
Travel Award (International Conference on Case-Based Reasoining)
Travel Award (AI Ethics & Society)
Rising Star Award (UCD Institute of Discovery)
Memberships
Associate member of Exeter College Oxford
Association of Computing Machinery
European Association of Algorithmic Fairness
Governance of Emerging Technology. Academic Collaborator.