NEXUS Workshop @ AAMAS 2026

For autonomous agents to be deployed safely and effectively, their decision-making must be robust and transparent for trustworthy behavior.

The opacity of current learning models is a major barrier to adoption where accountability is essential. NEXUS focuses on neurosymbolic reasoning as a foundation to building theory, applications, and tools for well-calibrated and trustworthy autonomy. This paradigm moves beyond a narrow focus on formal logic, concerning itself with the integration of deep and reinforcement learning algorithms with a broad spectrum of structured knowledge.

Program

9:00–10:10	Session 1
09:00–09:15	Workshop Opening
09:15–09:35	Robust Visual Planning under Partial Observability and Perceptual Uncertainty Guy Azran, Michael Navat, Sarah Keren
09:35–09:55	Towards Neuro-symbolic Causal Rule Synthesis, Verification, and Evaluation Grounded in Legal and Safety Principles Zainab Rehan, Christian Medeiros Adriano, Sona Ghahremani, Holger Giese
09:55–10:10	PSN Game: Game-theoretic Prediction and Planning via a Player Selection Network Tianyu Qiu, Eric Ouano, Fernando Palafox, Christian Ellis, David Fridovich-Keil
10:15–11:00	Coffee Break
11:00–12:30	Session 2
11:00–11:20	A Consensus-Bayesian Framework for Detecting Malicious Activity in Enterprise Directory Access Graphs Shilpa Noushad, Prayush Uppuluri, Sajan Kumar
11:20–11:30	Mini-break
11:30–12:30	Keynote talk – Francesca Toni (Imperial College London)
12:30–14:00	Lunch
14:00–15:30	Session 3
14:00–14:20	LLM-COGNICA: Towards Reliable Agent Decision Making Adamos Koumi, Antonis Kakas
14:20–14:40	Formal Specification Embeddings for Neuro-Symbolic Autonomy Beyazit Yalcinkaya, Marcell Vazquez-Chanlatte, Sanjit A. Seshia
14:40–14:50	Mini-break
14:50–15:10	Shape-Preserving Networks: Encoding Structural Knowledge through Derivative Constraints and Nested Integration Miguel Suau, Alec Edwards, Alex Durkin, Ian Howell, Giuseppe Pinto, Marc Artola, Daniel Jarne Ornia, Sayam Kumar
15:10–15:30	Probing the Linear Geometry of Deception in Neurosymbolic Negotiation Agents Teanna Sims
15:30–16:15	Coffee Break
16:15–17:40	Session 4
16:15–16:35	From a Neural-Symbolic Means to the Inductive-Instructional Loizos Michael
16:35–16:40	Mini-break
16:40–17:40	Panel: Neuro-Symbolic Reasoning at the Age of Multi-Modal LLMs C.C. (Cor) Steging, Michael Loizos, Andreas Theodorou
19:30–…	Dinner (TBC)

Important dates

17 Dec 2025	Call for Papers Release
27 Feb 2026	Submission Deadline
03 Apr 2026	Acceptance Notification
25 May 2026	Workshop

Call for papers

NEXUS aims to make progress in neurosymbolic reasoning for agents and multi-agent systems with a focus on topics around trustworthy AI. We invite submissions addressing urgent issues of efficiency, trust, and safety through neurosymbolic approaches, including but not limited to:

Causal inference Embedding causal graphs to enable reasoning about interventions and counterfactuals.

Compositionality & knowledge embedding Constructing complex behaviors from reusable modules and embedding symbolic knowledge.

Physics-informed learning Injecting differential equations or physical laws as side-information.

Strategic reasoning Game theory and mechanism design in multi-agent environments.

Optimization & planning Integrating classical optimizers, planners, or solvers within the learning loop.

Submission details

Submissions are handled via OpenReview through a single-round, single-blind review. While the platform enforces double-blind submissions, we do not require anonymity; authors should retain their affiliation and name on the paper. We welcome full papers (≤8 pages, AAMAS style) and shorter work-in-progress papers. Early-career researchers and students are particularly encouraged to present their ongoing research.

Organization

Georgios Bakirtzis Institut Polytechnique de Paris, France bakirtzis.net
bakirtzis@telecom-paris.fr

Andreas Theodorou Google Scholar
a.theodorou@bath.edu

David Fridovich-Keil UT Austin, USA dfridovi.github.io
dfk@utexas.edu

Jingqi Li UT Austin, USA https://sites.google.com/view/jingqi-li
jingqi.li@austin.utexas.edu

Antonis Kakas University of Cyprus, Cyprus http://www2.cs.ucy.ac.cy/~antonis
antonis@ucy.ac.cy

Jack McKinlay University of Bath, UK jmckinlay.com
jam218@bath.ac.uk

Program committee

Georgios Bakirtzis, Andreas Theodorou, Jingqi Li, Jack McKinlay, Franck Djeumou, Cody Fleming, Fabrizio Riguzzi, Vadim Malvone, Gioele Zardini, Ivan Ruchkin, Lasse Peters, Juan Carlos Nieves, Julian Padget, George Vouros, and Beyazit Yalcinkaya.