Reproduce “The Supervised IBP: Neighbourhood Preserving Infinite Latent Feature Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Novi Quadrianto, Viktoriia Sharmanska, David A. Knowles, Zoubin Ghahramani
The paper extends hash codes by learning extra features from neighbour/non-neighbour comparisons, using an Indian Buffet Process to infer how many features are active. In a synthetic example, added bits separate categories sharing a five-bit code. On Animals with Attributes, Super Probit IBP reaches 45.9 ± 4.1 versus 29.5 ± 3.2 for the hash baseline in the five-category, 30-nearest-neighbour setting; this is offline evidence, not a live-streaming result.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.