Compositional Semantics for Open Vocabulary Spatio-semantic Representations 事件

Compositional Semantics for Open Vocabulary Spatio-semantic Representations arXiv:2310.04981v2 Announce Type: replace Abstract: Vision-language models (VLMs) transform environment percepts into vision-language semantics interpretable by LLMs. However, completing complex tasks often requires reasoning about information beyond what is currently perceived. We propose latent compositional semantic embeddings z* as a principled learning-based knowledge representation for queryable spatio-semantic me