Generalized Planning Using Language Models and its Applications

Explore how large language models can revolutionize automated planning and decision-making systems in this 54-minute PhD proposal defense recording. Discover a comprehensive taxonomy of language model applications in planning, examine evaluations of both pretrained and fine-tuned models for plan generation, and learn about innovative neurosymbolic architectures that combine machine learning with symbolic reasoning to address current limitations. Investigate real-world implementations including collaborative information-retrieval assistants and manufacturing replanning systems that demonstrate the practical potential of learning-based planners for complex decision-making scenarios.