​​​​​​​

26. October 2026 - 28. October 2026

About the event

The 17th International Conference on Grammatical Inference

ICGI 2026 will take place at the TU Delft, the Netherlands, from 26 to 28 October 2026.

Grammatical Inference is the research area at the intersection of Machine Learning and Formal Language Theory. Since 1993, the International Conference on Grammatical Inference (ICGI) is the meeting place for presenting, discovering, and discussing the latest research results on the foundations of learning languages, from theoretical and algorithmic perspectives to their applications (natural language or document processing, bioinformatics, model checking and software verification, program synthesis, robotic planning and control, intrusion detection, etc).

Accepted papers

  • Automata Learning for Probabilistic IO Automata with Continuous Outputs by Benjamin von Berg and Bernhard Aichernig

  • A formal approach for understanding the behavior of constrained language models by Martín Iturbide, Sergio Yovine and Matías Carrasco

  • The Long and Short of It: An experimental investigation into the role of shorter strings in learning regular languages by Logan Swanson, Sarah Payne, Sicco Verwer, Sam van der Poel, Adil Soubki and Jeffrey Heinz.

  • Concise One-Layer Transformers Can Perform Dictionary Lookup by Anton Hampe, Lena Strobl, Robert Frank and Dana Angluin

  • Learnability of Grammars with Finite Exceptions by Jacob Rowen

  • Learning deterministic finite-state machines from the prefixes of a single string is NP-complete by Radu Cosmin Dumitru, Ryo Yoshinaka and Ayumi Shinohara

  • Compiling Search & Change Rules into Learnable Subsequential Finite State Transducers by Malek Azadegan

  • Linguistic Faithfulness Enables Decomposition Learning of Sequential Functions from Given Structure by William Schilling and David Mortensen

  • Learning DFA with neural networks: Self-Organising Automata by Aurélien Lemay

  • The Pumping Inference Algorithm: Exploiting the Pumping Lemma for CFGs to Learn PCFGs by Henrique Barros Lopes

  • Simulation Reductions for the Complexity of Subsequential Transducer Identification from Non-Characteristic Data by William Schilling

  • A Minimum Description Length Approach to Grammar Inference in Neural Networks by Matan Abudy, Orr Well, Emmanuel Chemla, Roni Katzir and Nur Lan

  • Identification in the Limit with Polynomial Time and Data of (k,l)-Local-Context Substitutable Grammars by Arnaud Patey and François Coste

  • Learning Bimachines from Aligned Examples by Mans Hulden and Michael Ginn

Beside the above-mentioned papers, there will be shorter presentations which discuss work-in-progress or already published work relevant to the ICGI community. These will be announced later.

Location

26. & 28. October - Aula TU Delft

Building 20

Mekelweg 5

2628 CC Delft

The Netherlands

27. October - Echo TU Delft

Building 29

Van Mourik Broekmanweg 5

2628 XE Delft

The Netherlands