active query selection

This work proposes a novel automated approach to adversarial query synthesis for database benchmarking by introducing Bayesian optimization—a technique previously unexplored in this context. Traditional database benchmarks rely heavily on manually crafted queries, which struggle to uncover high-difficulty test cases that stress system performance. The proposed method integrates automatic query generation with performance evaluation, substantially reducing the need for human intervention while efficiently producing queries that pose greater challenges to database systems. Experimental results demonstrate that the synthesized queries exhibit more than twice the optimization potential compared to those from existing benchmarks, significantly enhancing both the coverage and efficiency of database testing.

This paper addresses the challenge that incomplete and fragmented expert forecasts often yield aggregated predictions no better than random selection. To tackle this, we propose a novel robust prediction aggregation framework based on structured auxiliary queries. Methodologically, we design a game-theoretic incentive mechanism to elicit truthful belief reports from experts and systematically characterize the theoretical trade-off among query complexity, reasoning order, and aggregation error under an independent yet overlapping signals model. Our key contributions are: (i) the first precise characterization of the query–error trade-off—showing that, in the worst case, only *O*(*n*) queries suffice for optimal aggregation, with error decaying linearly in the number of queries; and (ii) establishing that when the effective number of experts exceeds √*n*, the aggregation error converges to zero—thereby breaking a fundamental performance bottleneck inherent in conventional aggregation methods.

This work addresses the problem of data sufficiency for linear optimization under cost vector uncertainty: identifying the minimal dataset that uniquely determines the optimal decision. Methodologically, it introduces the first geometric sufficiency criterion for linear programming, grounded in convex geometry and duality theory, to characterize the critical cost directions governing optimality; it further establishes a modeling framework for uncertainty sets and designs a task-driven data selection algorithm. Theoretically, it proves the existence of a small-scale, structured minimal cost dataset sufficient to fully recover the optimal solution. This work provides rigorous theoretical guarantees and an efficient constructive procedure for task-aware data acquisition, overcoming key limitations of conventional sufficiency analyses—namely, their reliance on statistical assumptions or large-sample requirements.

This work addresses robust overfitting in adversarial training, identifying excessive model confidence—i.e., high predictive certainty—on adversarial examples as a key cause of degraded robust generalization. To this end, we formally introduce the concept of *adversarial certainty*, defined as the variance of logits on adversarial samples, and establish its theoretical connection to robust generalization performance. Building upon this insight, we propose a generic robust optimization framework that actively reduces adversarial certainty during training without compromising clean classification discriminability. Extensive experiments across multiple image classification benchmarks demonstrate that our method significantly improves robust accuracy and effectively mitigates robust overfitting. These results empirically validate that explicitly controlling adversarial uncertainty is crucial for enhancing robust generalization.

In safety-critical applications, adversarial attacks can maliciously manipulate model uncertainty estimates—over- or under-estimating confidence—thereby undermining decision reliability and system usability. This work is the first to theoretically and empirically demonstrate that standard adversarial training methods (e.g., TRADES, PGD) inherently improve the robustness of uncertainty estimation, obviating the need for dedicated uncertainty-specific defenses. We conduct a systematic evaluation across multiple adversarially robust models on CIFAR-10 and ImageNet using the RobustBench benchmark, showing substantial gains in resilience against uncertainty-targeted attacks: uncertainty calibration error decreases by up to 42% compared to standard models. We further provide theoretical analysis proving that this robustness arises from an implicit regularization effect induced by adversarial perturbations on the confidence margin—effectively tightening the bounds of predictive confidence. Our findings bridge adversarial robustness and reliable uncertainty quantification, offering a principled, unified approach to trustworthy AI in high-stakes settings.

This work addresses the limited generalization of existing methods in complex scenarios by proposing a novel framework based on adaptive feature fusion and contrastive learning. The approach dynamically integrates multi-scale semantic information and incorporates a task-aware contrastive loss to enhance model robustness under distribution shifts. Extensive experiments demonstrate that the proposed method significantly outperforms state-of-the-art approaches across multiple benchmark datasets, achieving an average accuracy improvement of 3.2% while exhibiting superior cross-domain transferability. By offering a new perspective on improving visual model generalization, this study contributes a practical and effective solution, with code publicly released to facilitate future research.

This work addresses the vulnerability of traditional decision-oriented experimental design to unmodeled perturbations or hidden effects, which often leads to fragile decisions. It introduces adversarial robustness into Bayesian sequential experimental design for the first time, optimizing decision stability under uncertainty rather than nominal optimality by modeling worst-case perturbations from adversarial variables. The paper formalizes a robust optimality criterion for decision-making under such adversarial disturbances. By integrating Bayesian decision theory, adversarial robust optimization, and active learning, the proposed approach significantly enhances the stability and reliability of decisions on both synthetic and real-world scientific datasets, outperforming existing methods.

This work addresses the challenge of explanation stability in incremental information disclosure, where traditional explanations may be invalidated by subsequent updates. The authors introduce the notion of a “robust reason core”—a minimal set of features that remains sufficient for a prediction across all admissible future disclosures. By compiling defeasible classifiers into explicit boundary maps equipped with entry anchors and exit defeaters, the method enables efficient joint scanning of predictions, anchors, and defeater fronts. The study establishes the Σ₂^P-completeness of this problem for the first time and delineates a complete complexity landscape spanning P, coNP, NP, and Σ₂^P. Empirical evaluation on Boolean abstractions of standard datasets reveals that robust cores are remarkably small (single-digit size), enabling efficient exact auditing, while adversarial instances confirm the theoretical hardness, exhibiting core sizes of Θ(n).

This work addresses the challenge of applying traditional multi-winner voting rules in large-scale or attention-constrained settings, where eliciting complete preference rankings from voters is impractical. To overcome this limitation, the authors propose a structured-query framework for multi-winner elections that approximates an optimal committee by querying voters’ preferences over subsets of candidates within a limited budget. They formally define a cognitive cost function and axiomatic evaluation criteria, and introduce a query strategy based on recursively partitioning the candidate set. Experimental results demonstrate that this approach significantly outperforms alternative querying mechanisms across various election models and multi-winner rules—such as k-Borda—achieving high committee selection accuracy while substantially reducing the information acquisition cost.

This work addresses the often-overlooked degradation of predictive uncertainty quality caused by conventional adversarial training, which undermines selective classification performance despite improving model robustness. The study systematically reveals, for the first time, the adverse impact of adversarial training on uncertainty calibration and ranking. To mitigate this issue, the authors propose Evidence-based Adversarial Training (EV-AT), a novel approach grounded in evidential theory that jointly optimizes standard accuracy and uncertainty reliability in the Dirichlet parameter space. EV-AT employs an evidential loss combined with a robust evidential alignment loss to enforce consistency between predictions on clean and adversarial examples. Extensive experiments across multiple datasets and threat models demonstrate that EV-AT significantly outperforms existing methods, simultaneously enhancing both robust accuracy and selective classification performance, thereby advancing the Pareto frontier of the robustness–uncertainty trade-off.