Katarzyna (Kasia) Kobalczyk
Katarzyna (Kasia) Kobalczyk
Home
Publications
Contact
Light
Dark
Automatic
Active Task Disambiguation with LLMs
This paper formalizes task ambiguity in tasks specified in natural language and frames task disambiguation through Bayesian Experimental Design, leading to more effective strategies for LLMs to pose clarifying questions.
Kasia Kobalczyk
,
Nicolás Astorga
,
Tennison Liu
,
Mihaela van der Schaar
Jan 22, 2025
openreview
arXiv
PDF
code
Towards Automated Knowledge Integration From Human-Interpretable Representations
We introduce the paradigm of informed meta-learning, a novel approach to inductive bias specification based on human knowledge represented in any form, including unstructured natural language.
Kasia Kobalczyk
,
Mihaela van der Schaar
Jan 22, 2025
openreview
arXiv
PDF
code
Tabular Few-Shot Generalization Across Heterogenous Feature Spaces
We introduce new tabular few-shot learning solution capable of knowledge sharing between datasets with heterogenous sets of columns.
Max Zhu
,
Kasia Kobalczyk
,
Andrija Petrovic
,
Mladen Nikolic
,
Mihaela van der Schaar
,
Pietro Lio
,
Boris Delibašić
Sep 22, 2023
arXiv
PDF
cegpy: Modelling with chain event graphs in Python
Chain event graphs (CEGs) are a recent family of probabilistic graphical models that generalise the popular Bayesian networks (BNs) family. In this paper, we present cegpy—the first Python implementation of CEGs and the first across all languages to support structurally asymmetric processes.
Gareth Walley
,
Aditi Shenvi
,
Peter Strong
,
Kasia Kobalczyk
Aug 15, 2023
ScienceDirect
PDF
code
docs