Improving Predictive Inference under Covariate Shift by Weighting the Log-likelihood Function.¶
Shimodaira, H. (2000). Improving Predictive Inference under Covariate Shift by Weighting the Log-likelihood Function. Journal of Statistical Planning and Inference, 90(2), 227-244.
Cited by¶
3 citations across 3 artifacts.
Each citation links to the sentence it supports in the citing article.
Primes¶
- Training Serving Skew
- A model learns a predictor \(\hat{f}\) by minimising loss over a preparation environment whose inputs are drawn from distribution \(P_{\text{train}}(x)\), while the conditional relationship \(P(y \mid x)\) is assumed stable.
This sourceFormalizes covariate shift — input distribution differs between training and deployment while P(y|x) is stable.
- A model learns a predictor \(\hat{f}\) by minimising loss over a preparation environment whose inputs are drawn from distribution \(P_{\text{train}}(x)\), while the conditional relationship \(P(y \mid x)\) is assumed stable.
Domain-specific¶
- Covariate-Shift Blind Spot
- Covariate shift is the specific subcase of distribution shift in which P(X) moves while P(Y|X) is assumed fixed
This sourceThe definition of covariate shift as the case where the input distribution P(X) changes between training and deployment while the conditional P(Y|X) is assumed unchanged.
- Covariate shift is the specific subcase of distribution shift in which P(X) moves while P(Y|X) is assumed fixed
- Gold-Standard Erosion
- Gold-standard erosion is a change in the reference label, not the input-outcome relationship; the world and the model can be unchanged while only the answer key moves, and retraining against that key recovers nothing. Not covariate or dataset shift. Covariate shift is a change in the input distribution — the deployment population moved — remedied by importance weighting (reweighting the training distribution), the correction the covariate-shift literature gives.
This sourceImproves predictive inference under covariate shift by weighting the log-likelihood function.
Supported in partVerified against the source
- Gold-standard erosion is a change in the reference label, not the input-outcome relationship; the world and the model can be unchanged while only the answer key moves, and retraining against that key recovers nothing. Not covariate or dataset shift. Covariate shift is a change in the input distribution — the deployment population moved — remedied by importance weighting (reweighting the training distribution), the correction the covariate-shift literature gives.
Verification¶
Does it exist? Confirmed. This work's DOI resolves to a registered record, which fixes its identity. That is all it fixes.
Does it back the claim? Read against the text for 1 of 3 citations: 1 supported in part. Each verdict is shown under its citation below, with what in the work backs the sentence.
Support is checked per citation rather than per work — the same source can be cited soundly in one article and wrongly in another. Per-citation recording began recently, so a citation with no recorded check is a gap in the record rather than evidence it went unchecked.
See how references were verified.
Registry ID ref:ac78addb840f · see in the full table