Date Posted: 12/23/2014

Graduating PhD student Anshumali Shrivastava, in work with Ping Li, received one of two best paper awards at NIPS 2014 for "Asymmetric LSH (ALSH) for Sublinear Time Maximum Inner Product Search (MIPS)".  This is out of 414 accepted papers from 1678 submissions.

One of the reviews (NIPS publishes reviews for accepted papers) states, "The result is very exciting, as it solves an important open problem, creating a method for fast hashing-based retrieval for an important class of similarities, that previously could only approximate dot product via cosine distances via Charikar's method. The detailed analysis showing viability and practical recipes for parameter settings ... is excellent, and provides a ready-to-use recipe to practitioners."