The paper introduces a method for maximizing the area under the receiver operating characteristic curve (AUC) when only biased positive and unlabeled (PU) data are available. It leverages confidence scores—probabilities that an instance is positive—associated with a small set of labeled positives to derive an AUC risk estimator that accounts for bias. Experiments on eight real-world datasets demonstrate the method’s effectiveness.
arXiv:2607. 09816v1 Announce Type: new Abstract: Class imbalance poses a fundamental challenge in risk-sensitive applications such as fraud detection and medical diagnosis, where minority-class samples are scarce yet critical for accurate classification.
The paper introduces soft‑label‑based estimators for the Bayes‑optimal balanced error rate (BER) and area under the ROC curve (AUC), extending from a clean setting with known class priors to a realistic scenario with unknown priors and corrupted soft labels. It also adapts the FeeBee evaluation framework to assess these estimators without needing the true optimum, providing practical evaluation scores for any estimator of optimal BER or AUC. Experiments on synthetic and real datasets confirm the effectiveness of both the estimators and the evaluation method.
arXiv:2607. 13428v1 Announce Type: new Abstract: Positive-Unlabeled (PU) learning aims to achieve high-accuracy binary classification with limited labeled positive examples and numerous unlabeled ones.
The paper demonstrates that common binary classification metrics—Matthews' correlation coefficient, Cohen's κ, the F-score, and the Jaccard similarity—are not robust to extreme class imbalance, as the Bayes classifier’s true positive rate tends to zero when the minority class proportion vanishes. To address this, the authors propose robustified versions of these metrics that include a tuning parameter, ensuring that the Bayes-optimal classifier’s threshold remains bounded and its true positive rate stays above zero even in highly imbalanced scenarios. The study provides theoretical bounds, simulation results, and practical guidance on applying these robust metrics to real data, such as a credit‑default dataset, and discusses their relationship to ROC and precision‑recall curves.
arXiv:2605. 03289v2 Announce Type: replace-cross Abstract: Detecting observations from a minority class under severe class imbalance is a central challenge in applications such as fraud detection, medical screening, and industrial quality control.
arXiv:2407. 05370v3 Announce Type: replace Abstract: Semi-supervised learning (SSL) algorithms often struggle to perform well when trained on imbalanced data.
arXiv:2607. 11947v1 Announce Type: cross Abstract: Typical semi-supervised learning (SSL) methods rely on distributional assumptions, and their performance degrades when these are violated.
arXiv:2209. 01754v5 Announce Type: replace-cross Abstract: The empirical risk minimization approach to data-driven decision making requires access to training data drawn under the same conditions as those that will be faced when the decision rule is deployed.
arXiv:2608.30699v1 Announce Type: cross Abstract: Long-tailed distributions are prevalent in real-world semi-supervised learning (SSL), where pseudo-labels tend to favor majority classes, leading to...
SAGE (Subpopulation-Aware Generative Enhancement) is a two-stage generative augmentation framework designed to mitigate spurious correlations in machine learning when group labels are unavailable. It uses cluster-derived sub-labels and class labels to fine‑tune a conditional generative model and text encoder, producing synthetic data that fills underrepresented regions and creates a balanced validation set for last‑layer reweighting. Experiments show SAGE improves worst‑group accuracy to 89.5%, 85.7%, and 79.1% on Waterbirds, CelebA, and MetaShift, outperforming existing group‑label‑free baselines by up to 7.7 percentage points.
arXiv:2608. 11162v1 Announce Type: new Abstract: The Naive Bayes (NB) classifier remains a standard choice for categorical data, yet its widely used smoothing rules, such as Laplace, Lidstone, Krichevsky-Trofimov, and the $m$-estimate, all prescribe a fixed smoothing strength that ignores feature cardinality, sample size, and class imbalance, inducing a non-vanishing bias on modern high-cardinality tabular data.
The paper introduces the concept of observational multiplicity, where multiple probabilistic classifiers can perform similarly yet produce conflicting predictions, undermining interpretability and safety. It proposes measuring this arbitrariness through a regret metric that captures how predictions could shift with different training labels. The authors present a general method to estimate regret, show it varies across dataset groups, and discuss its use for safety via abstention and targeted data collection.