Okapi: generalising better by making statistical matches match

Bartlett, Myles, Romiti, Sara, Sharmanska, Viktoriia and Quadrianto, Novi (2022) Okapi: generalising better by making statistical matches match. Thirty-sixth Conference on Neural Information Processing Systems, New Orleans, November 28th through December 9th. Published in: Advances in Neural Information Processing Systems 36. NeurIPS (Accepted)

[img] PDF - Accepted Version
Restricted to SRO admin only

Download (2MB)
Official URL: https://papers.nips.cc/


We propose Okapi, a simple, efficient, and general method for robust semi-supervised learning based on online statistical matching. Our method uses a nearest-neighbours-based matching procedure to generate cross-domain views for a consistency loss, while eliminating statistical outliers. In order to perform the online matching in a runtime- and memory-efficient way, we draw upon the self-supervised literature and combine a memory bank with a slow-moving momentum encoder. The consistency loss is applied within the feature space, rather than on the predictive distribution, making the method agnostic to both the modality and the task in question. We experiment on the WILDS 2.0 datasets (Sagawa et al., 2022), which significantly expands the range of modalities, applications, and shifts available for studying and benchmarking real-world unsupervised adaptation. Contrary to Sagawa et al., 2022, we show that it is in fact possible to leverage additional unlabelled data to improve upon empirical risk minimisation (ERM) results with the right method. Our method outperforms the baseline methods in terms of out-of-distribution (OOD) generalisation on the iWildCam (a multi-class classification task) and PovertyMap (a regression task) image datasets as well as the CivilComments (a binary classification task) text dataset. Furthermore, from a qualitative perspective, we show the matches obtained from the learned encoder are strongly semantically related. Code for our paper is publicly available at https://github.com/wearepal/okapi/.

Item Type: Conference Proceedings
Schools and Departments: School of Engineering and Informatics > Informatics
Depositing User: Novi Quadrianto
Date Deposited: 12 Oct 2022 08:48
Last Modified: 19 Oct 2022 09:58
URI: http://sro.sussex.ac.uk/id/eprint/108435

View download statistics for this item

📧 Request an update
Project NameSussex Project NumberFunderFunder Ref
BayesianGDPR - Bayesian Models and Algorithms for Fairness and TransparencyG2903European Union10.3030/851538