Foundation Models for Decision Making

We invite submissions in the area of decision making on topics such as, but not limited to:

• Applying foundation models (e.g., autoregressive LMs, diffusion models, masked autoencoders, contrastive pretraining, prompting of large-pretrained models) to decision making (e.g., control, planning, online / offline RL, imitation learning).

• Learning multi-modal, multi-task, multi-environment, and generalist policies.

• Theoretical understanding of the roles foundation models play in decision making.

• Text and image models that involve (human-in-the-loop) interactions.

• Goal-driven reasoning under the formalism of decision making.

• Evaluation protocols, benchmarks, and datasets for applying foundation models to decision making.

• Practical applications of large-scale decision making and how such application guide novel problem formulations.

• New and open problems in decision making that could benefit from foundation models.

Please visit the front page for a more detailed summary of what we are excited about. We welcome both theoretical and empirical contributions. Submissions can range from 4 - 9 pages in NeurIPS or ICLR format, where shorter submissions with proof-of-concept demonstrations are welcome.