Doubly Robust Deep Kernel Tests for Structured Outcomes
The paper proposes a test to determine if two interventions induce the same outcome distribution in high-dimensional data.
The paper introduces a new test called DR-KTE-Deep that uses mean embeddings and learned deep features to compare intervention outcomes.
Before reading this…
Applications
- →Comparing the effects of different medical treatments
- →Evaluating the impact of different marketing strategies
To understand this paper, make sure you know these concepts first:
- Knowledge of machine learning and deep learningfind papers →
- Understanding of high-dimensional datafind papers →
Abstract
More Like ThisWe propose a test of whether two interventions induce the same outcome distribution when outcomes are structured and high-dimensional, such as images, text and graphs. We represent the interventional distributions as mean embeddings on learned deep features, such that the treatment effect is a doubly robust representation of the difference in mean embeddings. Our test, DR-KTE-Deep, optimises the learned features so as to maximise an estimated power criterion on this treatment effect representation. To motivate our learning objective, we show that the power criterion characterises the asymptotic signal-to-noise ratio of the test under fixed alternatives, and we derive a uniform finite-sample bound for its regularised estimate. We show that, with the learned kernel fixed, a held-out test is asymptotically valid under the null given suitable nuisance-estimation rates. Empirically, DR-KTE-Deep achieves substantially higher power than fixed and bandwidth-optimised kernel baselines across image, text and graph outcomes, while maintaining well-calibrated Type-I error.