Andrew McCallum

Machine-learning / NLP researcher (UMass Amherst, IESL) recurring across the region-embedding line in this vault:

• Gaussian Embedding (Vilnis & McCallum, ICLR 2015) — words as Gaussians. See Gaussian Embedding.
• Answering Compositional Queries with Set-Theoretic Embeddings (Dasgupta, McCallum, Rendle, Zhang, 2023) — box embeddings for compositional item–attribute queries.

His group is associated with much of the probabilistic box-embedding research that produced Box Embedding and Word2Box.

Key Contributions

• Gaussian Embedding
• Box Embedding / Set-Theoretic Embeddings

Articles

• Answering Compositional Queries with Set-Theoretic Embeddings

Related People

• Luke Vilnis — Gaussian Embedding & Box Lattice co-author
• Shib Sankar Dasgupta — Word2Box, Gumbel Box, compositional queries