[CFP] Unifying Concept Representation Learning workshop @ ICLR 2026 - deadline Jan 30
Sara Magliacane
ICLR 2026 workshop on Unifying Concept Representation Learning
https://ucrl-iclr26.github.io/
Several areas at the forefront of AI research are currently witnessing a convergence of interests around the problem of learning high-quality concepts from data. Concepts have become a central topic of study in neuro-symbolic integration (NeSy). NeSy approaches integrate perception – usually implemented by a neural backbone – and symbolic reasoning by employing concepts to glue together these two steps: the latter relies on the concepts detected by the former to produce suitable outputs. Concepts are also used in Explainable AI (XAI) by recent post-hoc explainers and self-explainable architectures as a building block for constructing high-level justifications of model behavior. Compared to, e.g., saliency maps, these can portray a more abstract and understandable picture of the machine’s reasoning process, potentially improving interpretability, interactivity, and trustworthiness, to the point that concepts have been called the lingua franca of human-AI interaction.
NeSy and XAI methods hinge on learned concepts being “high-quality”. Concepts with misaligned semantics may compromise the meaning of model explanations, out-of-distribution behavior of NeSy architectures and human understanding of the underlying systems. Recent works propose to leverage disentangled representations to mitigate concept leakage, i.e., the presence of irrelevant information in the learned concepts. Causal Representation Learning (CRL) is a generalization of disentangled representation learning, when the latent variables are dependent on each other, e.g., due to causal relations.
The potential of leveraging CRL to learn more robust and leak-proof concepts is an emerging area of research with a growing number of approaches, but many open questions remain. In particular, what properties high-quality concepts should satisfy is unclear. Moreover, despite studying the same underlying object, research in NeSy, XAI and CRL is proceeding on mostly independent tracks, with minimal knowledge transfer. Separate branches differ in their working definitions of what concepts are and what desiderata they ought to satisfy, on what data and algorithms they should be learned with, and on how to properly assess their quality. This also means that approaches in one area often ignore insights from the others. As a result, the central issue of how to properly learn and evaluate concepts is largely unanswered.
The aim of this ICLR 2026 workshop is to bring together researchers from NeSy, XAI and CRL and from both industry and academia, who are interested in learning robust, semantically meaningful concepts. We welcome submissions on the following topics:
Foundations of concept representations and learning in XAI, CRL and NeSy.
Supervised and unsupervised techniques for learning concepts from observational and interventional data, raw inputs, and pre-trained embeddings.
Techniques for learning concepts in non-standard settings, e.g., causal abstraction.
Design and evaluation of concept-based XAI techniques and self-explainable concept-based models.
Interactive human-machine concept acquisition and alignment.
Applications of concept-based AI systems, including but not limited to, reasoning, causality, formal verification, interactive learning, and explainability.
Metrics and evaluation techniques for assessing the quality of learned concepts, with a focus on down-stream applications.
Important Dates & Details
Paper submission deadline: January 30th, 2026 23:59 AoE
Notification to authors: March 1st, 2026 23:59 AoE
Workshop date: April 26 or 27, 2026
Workshop location: Rio de Janeiro, Brazil
We invite submissions on on-going research that have not yet been published in a venue with proceedings. While we welcome unfinished work, submissions in this track should contain original ideas, new connections between research fields, or novel results. Submissions should be formatted using the ICML latex template and formatting instructions. Papers should be up to 6 pages in length, including all main results, figures, and tables. Appendices containing additional details are allowed, but reviewers are not expected to take this into account.
Submission Link: https://openreview.net/group?id=ICLR.cc/2026/Workshop/UCRL#tab-your-consoles
Organization committee
Amit Dhurandhar (IBM Research)
Amir-Hossein Karimi (U Waterloo)
Sara Magliacane (U Amsterdam)
Stefano Teso (U Trento)
Efthymia Tsamoura (Huawei Research)
Zhe Zeng (U Virginia)