English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

This work studies the problem of learning unbiased algorithms from biasedfeedback for recommendation. We address this problem from a novel distributionshift perspective. Recent works in unbiased recommendation have advanced thestate-of-the-art with various techniques such as re-weighting, multi-tasklearning, and meta-learning. Despite their empirical successes, most of themlack theoretical guarantees, forming non-negligible gaps between theories andrecent algorithms. In this paper, we propose a theoretical understanding of whyexisting unbiased learning objectives work for unbiased recommendation. Weestablish a close connection between unbiased recommendation and distributionshift, which shows that existing unbiased learning objectives implicitly alignbiased training and unbiased test distributions. Built upon this connection, wedevelop two generalization bounds for existing unbiased learning methods andanalyze their learning behavior. Besides, as a result of the distributionshift, we further propose a principled framework, Adversarial Self-Training(AST), for unbiased recommendation. Extensive experiments on real-world andsemi-synthetic datasets demonstrate the effectiveness of AST.

Reconsidering Learning Objectives in Unbiased Recommendation with Unobserved Confounders