The 1st International Workshop on Neuro-Symbolic Debates for Safeguarded Generative AI (NeSyDebates 2026)
Overview
Recent advances in large language models (LLMs) and foundation models have transformed artificial intelligence, enabling impressive capabilities in reasoning, generation, and decision-making. Yet these systems still face critical challenges, including hallucinations, unreliable reasoning, factual inconsistencies, vulnerability to adversarial inputs, and limited explainability. As generative AI is deployed in high-stakes domains such as law, healthcare, finance, and public administration, ensuring safety, trustworthiness, and accountability has become an urgent research priority.
Neuro-symbolic AI offers a promising path by combining the learning capabilities of neural networks with the rigor, transparency, and verifiability of symbolic reasoning. In parallel, debate-based AI frameworks—in which multiple agents reason, critique, and verify one another collaboratively or competitively—have emerged as a powerful approach to improving reasoning quality and reducing errors.
The Neuro-Symbolic Debates for Safeguarded Generative AI (NeSyDebates) workshop will bring together researchers in machine learning, natural language processing, knowledge representation, automated reasoning, multi-agent systems, human-AI interaction, and AI safety. Together, participants will explore how neuro-symbolic methods and debate-based reasoning can be integrated to build trustworthy, explainable, and safeguarded generative AI systems.
The workshop welcomes theoretical advances and practical applications and seeks to foster interdisciplinary collaboration across academia, industry, and government.
Why This Workshop?
NeSyDebates focuses on the intersection of neural learning, symbolic reasoning, and structured debate. By connecting these research communities, the workshop aims to advance generative AI systems that are safer, more reliable, more transparent, and easier to verify.
Topics of Interest
Neuro-Symbolic AI
- Neuro-symbolic learning and reasoning
- Neural theorem proving
- Logical reasoning with foundation models
- Symbolic planning and knowledge-guided generation
- Knowledge graph reasoning
- Rule induction and program synthesis
- Differentiable reasoning
- Formal verification for AI systems
Debate-Based AI
- Multi-agent debate frameworks
- Self-debate and collaborative reasoning
- Argumentation-based AI
- Multi-agent verification
- Debate protocols for LLMs
- Consensus and disagreement modeling
- Explainable debate systems
- Interactive reasoning
Safeguarded Generative AI
- AI safety, alignment, and trustworthy AI
- Hallucination detection and mitigation
- Factuality verification
- Uncertainty estimation
- Robustness to adversarial attacks
- Safe LLM inference
- Constitutional AI
- Alignment techniques
- Explainable LLMs
- Responsible AI
Learning and Inference
- Test-time reasoning
- Speculative decoding
- Retrieval-augmented generation (RAG)
- Agentic AI
- Long-context reasoning
- Continual learning
- Memory-augmented models
- Efficient inference
- Knowledge editing and unlearning
Domain Applications
- Legal AI
- Healthcare AI
- Scientific AI
- Finance
- Robotics
- Cybersecurity
- Government AI
- Education
- Industrial AI
- Physical AI
Resources
- Benchmarks for trustworthy AI
- Evaluation metrics
- Datasets
- Shared tasks
- Reproducibility
- Open-source systems
Paper Submission
We invite original research papers, novel algorithms and theoretical contributions, system demonstrations, benchmark and dataset papers, position and vision papers, industrial case studies, and well-motivated negative results. Submissions must describe unpublished work and will undergo double-blind peer review.
Submission Format: An 8-page (max) PDF manuscript conforming to IEEE template (Latex, Word).
Submission Link: https://cmt3.research.microsoft.com/KSE2026 (Track: Workshop – NeSyDebates 2026)
Important Dates (Tentative)
| Milestone | Date |
| Paper submission | September 5, 2026 |
| Notification of acceptance | September 20, 2026 |
| Camera-ready papers | October 10, 2026 |
| Workshop | November 10, 2026 |
Proceedings
Accepted papers will appear in the workshop proceedings, subject to the host conference’s publication policy. Selected authors may be invited to submit extended versions to a special issue of an international journal.
Invited Speakers (Tentative)
Invited speakers will be announced. The workshop plans to feature leading researchers from academia and industry working in neuro-symbolic AI, AI safety, large language models, automated reasoning, argumentation, multi-agent systems, and trustworthy AI.
Organizers
| Organizer | Affiliation |
| Nguyen Le Minh (Co-chair) | Japan Advanced Institute of Science and Technology, Japan |
| Ken Satoh (Co-chair) | National Institute of Informatics, Japan |
| Vu Tran | Japan Advanced Institute of Science and Technology, Japan |
| Trung Vo | Japan Advanced Institute of Science and Technology, Japan |
| May Myo Zin | National Institute of Informatics, Japan |
| Wachara Fungwacharakorn | National Institute of Informatics, Japan |
| Kong Yuntao | National Institute of Informatics, Japan |
| Francesca Toni (Co-chair) | Imperial College London, United Kingdom |
| Felix Steffek (Co-chair) | University of Cambridge, United Kingdom |
| Soteris Demetriou (Co-chair) | Imperial College London, United Kingdom |
| Alessandra Russo (Co-chair) | Imperial College London, United Kingdom |
| SUMIDA Mihoko (Co-chair) | Hitotsubashi University, Japan |
| Nguyen Minh Phuong | Japan Advanced Institute of Science and Technology, Japan |
| Kieu Que Anh | Japan Advanced Institute of Science and Technology, Japan |
| Le Nguyen Khang | Japan Advanced Institute of Science and Technology, Japan |
| Nguyen Dieu Hien | Japan Advanced Institute of Science and Technology, Japan |
Contact
nguyenml@jaist.ac.jp