Research

Below is a list of recent papers representative of on-going research themes. You can find a complete list of my publications on my Google Scholar profile or in my CV.

Transfer learning

This work focuses on the development of theory and methods for increasing the robustness of probabilistic machine learning methods to distribution shift.

Sloman, S.J., Caprio, M., & Kaski, S. (2026). Epistemic Errors of Imperfect Multitask Learners When Distributions Shift. arXiv preprint. doi:10.48550/arXiv.2505.23496

Sloman, S.J., Martinelli, J., & Kaski, S. (2025). Proxy-informed Bayesian transfer learning with unknown sources. 41st conference on Uncertainty in Artificial Intelligence. url

Bayesian experimental design

My work has attempted to understand the role of misspecification in Bayesian experimental design.

Tang, R., Sloman, S.J., & Kaski, S. (2026). Representative, Informative, and De-Amplifying: Requirements for Robust Bayesian Active Learning under Model Misspecification. Twenty-Ninth Annual Conference on Artificial Intelligence and Statistics. url

Sloman, S.J., Cavagnaro, D.R., & Broomell, S.B. (2024). Knowing what to know: Implications of the choice of prior distribution on the behavior of adaptive design optimization. Behavior Research Methods. doi:10.3758/s13428-024-02410-7

Sloman, S.J., Bharti, A., Martinelli, J., & Kaski, S. (2024). Bayesian Active Learning in the Presence of Nuisance Parameters. 40th conference on Uncertainty in Artificial Intelligence [oral presentation]. url

Sloman, S.J., Oppenheimer, D.M., Broomell, S.B., & Shalizi, C.R. (2022). Characterizing the robustness of Bayesian adaptive experimental designs to active learning bias. arXiv preprint. doi:10.48550/arXiv.2205.13698

Model parsimony

Machine learning has achieved remarkable success with overparameterised models. This work explores the implications of this for scientific theory and practice, especially in contexts in which "simpler" models have traditionally been preferred.

Dubova, M.* & Sloman, S.J.* (*equal contribution) (2026). Excess Capacity Learning. Accepted in Behavioral and Brain Sciences. url

Dubova, M., Chandramouli, S., Gigerenzer, G., Grünwald, P., Holmes, W., Lombrozo, T., Marelli, M., Musslick, S., Nicenboim, B., Ross, L., Shiffrin, R., White, M., Wagenmakers, E-J.*, Bürkner, P-C.*, Sloman, S.J.* (*joint senior authors) (2025). Is Ockham's razor losing its edge? New perspectives on the principle of model parsimony. PNAS. doi:10.1073/pnas.2401230121