Exploiting geometric structure in mixture proportion estimation with generalised Blanchard-Lee-Scott estimators
Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
Authors
Colleges, School and Institutes
Abstract
Mixture proportion estimation is a building block in many weakly supervised classification tasks (missing labels, label noise, anomaly detection). Estimators with finite sample guarantees help analyse algorithms for such tasks, but so far only exist for Euclidean and Hilbert space data. We generalise the framework of Blanchard, Lee and Scott to allow extensions to other data types, and exemplify its use by deducing novel estimators for metric space data, and for randomly compressed Euclidean data – both of which make use of favourable geometry to tighten guarantees. Finally we demonstrate a theoretical link with the state of the art estimator specialised for Hilbert space data.
Details
Original language | English |
---|---|
Title of host publication | 30th International Conference on Algorithmic Learning Theory (ALT'19) |
Publication status | Published - 2019 |
Event | 30th International Conference on Algorithmic Learning Theory (ALT'19) - Chicago, United States Duration: 22 Mar 2019 → 24 Mar 2019 |
Publication series
Name | Proceedings of Machine Learning Research |
---|---|
Volume | 98 |
ISSN (Electronic) | 2640-3498 |
Conference
Conference | 30th International Conference on Algorithmic Learning Theory (ALT'19) |
---|---|
Country | United States |
City | Chicago |
Period | 22/03/19 → 24/03/19 |
Keywords
- Mixture proportion estimation, metric spaces, covering dimension, randonm projections, Gaussian width