Reach-Avoid Derived (RAD) Embeddings

Latent Temporal Task Representations for Multi-Task Policy Learning

Publications

Learning Task-Conditioned Policies with Natural Language and Graph Embeddings of Formal Specifications

Beyazit Yalcinkaya, Marcell Vazquez-Chanlatte, and Sanjit A. Seshia

3rd International Conference on Neuro-symbolic Systems (NeuS) 2026 (to appear)

[Paper] [Code]

Automata-Conditioned Cooperative Multi-Agent Reinforcement Learning

Beyazit Yalcinkaya, Marcell Vazquez-Chanlatte, Ameesh Shah, Hanna Krasowski, and Sanjit A. Seshia

Forty-Third International Conference on Machine Learning (ICML) 2026 (to appear)

[Paper] [Code]

Provably Correct Automata Embeddings for Optimal Automata-Conditioned Reinforcement Learning

Beyazit Yalcinkaya, Niklas Lauffer, Marcell Vazquez-Chanlatte, and Sanjit A. Seshia

2nd International Conference on Neuro-symbolic Systems (NeuS) 2025

[Paper] [Code]

Compositional Automata Embeddings for Goal-Conditioned Reinforcement Learning

Beyazit Yalcinkaya*, Niklas Lauffer*, Marcell Vazquez-Chanlatte*, and Sanjit A. Seshia (*equal contribution)

The Thirty-Eighth Annual Conference on Neural Information Processing Systems (NeurIPS) 2024

[Paper] [Code] [Webpage]

Automata-Conditioned Reinforcement Learning with Experience Replay

Beyazit Yalcinkaya*, Niklas Lauffer*, Marcell Vazquez-Chanlatte, and Sanjit A. Seshia (*equal contribution)

NeurIPS 2023 Workshop on Goal-Conditioned Reinforcement Learning 2023

[Paper] [Code]

Get Involved

If you are interested in collaborating or learning more, feel free to reach out or check the project repository.