Estimating calibration error under label shift without labels

• URL: http://arxiv.org/abs/2312.08586v1
• Date: Thu, 14 Dec 2023 01:18:51 GMT
• Title: Estimating calibration error under label shift without labels
• Authors: Teodora Popordanoska, Gorjan Radevski, Tinne Tuytelaars, Matthew B. Blaschko
• Abstract summary: Existing CE estimators assume access to labels from the target domain, which are often unavailable in practice, i.e., when the model is deployed and used. This work proposes a novel CE estimator under label shift, which is characterized by changes in the marginal label distribution $p(Y)$ while keeping the conditional $p(X|Y)$ constant between the source and target distributions. Our contribution is an approach, which, by leveraging importance re-weighting of the labeled source distribution, provides consistent and unbiased CE estimation with respect to the shifted target distribution.
• Score: 47.57286245320775
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: In the face of dataset shift, model calibration plays a pivotal role in ensuring the reliability of machine learning systems. Calibration error (CE) is an indicator of the alignment between the predicted probabilities and the classifier accuracy. While prior works have delved into the implications of dataset shift on calibration, existing CE estimators assume access to labels from the target domain, which are often unavailable in practice, i.e., when the model is deployed and used. This work addresses such challenging scenario, and proposes a novel CE estimator under label shift, which is characterized by changes in the marginal label distribution $p(Y)$, while keeping the conditional $p(X|Y)$ constant between the source and target distributions. Our contribution is an approach, which, by leveraging importance re-weighting of the labeled source distribution, provides consistent and asymptotically unbiased CE estimation with respect to the shifted target distribution. Empirical results across diverse real-world datasets, under various conditions and label-shift intensities, demonstrate the effectiveness and reliability of the proposed estimator.

Related papers

• Beyond Calibration: Assessing the Probabilistic Fit of Neural Regressors via Conditional Congruence [2.2359781747539396]
  Deep networks often suffer from overconfidence and misaligned predictive distributions. We introduce a metric, Conditional Congruence Error (CCE), that uses conditional kernel mean embeddings to estimate the distance between the learned predictive distribution and the empirical, conditional distribution in a dataset. We show that using to measure congruence 1) accurately quantifies misalignment between distributions when the data generating process is known, 2) effectively scales to real-world, high dimensional image regression tasks, and 3) can be used to gauge model reliability on unseen instances.
  arXiv  Detail & Related papers  (2024-05-20T23:30:07Z)
• Cal-SFDA: Source-Free Domain-adaptive Semantic Segmentation with Differentiable Expected Calibration Error [50.86671887712424]
  The prevalence of domain adaptive semantic segmentation has prompted concerns regarding source domain data leakage. To circumvent the requirement for source data, source-free domain adaptation has emerged as a viable solution. We propose a novel calibration-guided source-free domain adaptive semantic segmentation framework.
  arXiv  Detail & Related papers  (2023-08-06T03:28:34Z)
• ELSA: Efficient Label Shift Adaptation through the Lens of Semiparametric Models [2.73424570009809]
  Under the label shift context, the marginal distribution of the label varies across the training and testing datasets. We propose a moment-matching framework for adapting the label shift based on the geometry of the influence function. Under such a framework, we propose a novel method named underlineEfficient underlineLabel underlineShift underlineAdaptation (ELSA)
  arXiv  Detail & Related papers  (2023-05-30T15:31:44Z)
• Predicting Class Distribution Shift for Reliable Domain Adaptive Object Detection [2.5193191501662144]
  Unsupervised Domain Adaptive Object Detection (UDA-OD) uses unlabelled data to improve the reliability of robotic vision systems in open-world environments. Previous approaches to UDA-OD based on self-training have been effective in overcoming changes in the general appearance of images. We propose a framework for explicitly addressing class distribution shift to improve pseudo-label reliability in self-training.
  arXiv  Detail & Related papers  (2023-02-13T00:46:34Z)
• Leveraging Unlabeled Data to Predict Out-of-Distribution Performance [63.740181251997306]
  Real-world machine learning deployments are characterized by mismatches between the source (training) and target (test) distributions. In this work, we investigate methods for predicting the target domain accuracy using only labeled source data and unlabeled target data. We propose Average Thresholded Confidence (ATC), a practical method that learns a threshold on the model's confidence, predicting accuracy as the fraction of unlabeled examples.
  arXiv  Detail & Related papers  (2022-01-11T23:01:12Z)
• Predicting with Confidence on Unseen Distributions [90.68414180153897]
  We connect domain adaptation and predictive uncertainty literature to predict model accuracy on challenging unseen distributions. We find that the difference of confidences (DoC) of a classifier's predictions successfully estimates the classifier's performance change over a variety of shifts. We specifically investigate the distinction between synthetic and natural distribution shifts and observe that despite its simplicity DoC consistently outperforms other quantifications of distributional difference.
  arXiv  Detail & Related papers  (2021-07-07T15:50:18Z)
• Distribution-free uncertainty quantification for classification under label shift [105.27463615756733]
  We focus on uncertainty quantification (UQ) for classification problems via two avenues. We first argue that label shift hurts UQ, by showing degradation in coverage and calibration. We examine these techniques theoretically in a distribution-free framework and demonstrate their excellent practical performance.
  arXiv  Detail & Related papers  (2021-03-04T20:51:03Z)
• Transferable Calibration with Lower Bias and Variance in Domain Adaptation [139.4332115349543]
  Domain Adaptation (DA) enables transferring a learning machine from a labeled source domain to an unlabeled target one. How to estimate the predictive uncertainty of DA models is vital for decision-making in safety-critical scenarios. TransCal can be easily applied to recalibrate existing DA methods.
  arXiv  Detail & Related papers  (2020-07-16T11:09:36Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.