confidence-based action gating

This work proposes a novel representation learning framework that addresses the limited representational capacity of existing methods in complex scenes by integrating adaptive multi-scale fusion with contrastive learning. The approach dynamically aggregates multi-level features and incorporates a structure-aware contrastive loss, thereby enhancing the model’s ability to jointly capture fine-grained semantics and global contextual information. Extensive experiments demonstrate that the proposed framework consistently outperforms state-of-the-art methods across multiple benchmark datasets, achieving substantial improvements in both accuracy and robustness. These results establish a promising new direction for unsupervised and semi-supervised representation learning.

This work addresses the limitation of existing safety-critical systems, which typically evaluate only predictive accuracy while lacking rigorous validation of the overall calibration of predicted probability distributions. To bridge this gap, the authors propose a modular calibration testing framework that decouples the calibration process into four interchangeable components: data model, scoring rule, hypothesis formulation, and statistical test procedure. Built upon formal statistical hypothesis testing, the framework provides a single accept/reject decision for the entire predictive distribution. Crucially, it rejects only overly confident predictions while tolerating reasonable deviations, thereby balancing practicality with flexibility. Empirical evaluations on weather forecasting and robotic pose estimation tasks demonstrate that the framework effectively supports reliable deployment in safety-critical applications.

This work addresses the calibration bias in tool-augmented agents during multi-turn tasks, where agents exhibit overconfidence when using noisy evidence-providing tools (e.g., web search) but better calibration with verification-oriented tools (e.g., code interpreters). The study is the first to reveal this dichotomy between tool types and confidence calibration. To mitigate this issue, the authors propose a reinforcement learning fine-tuning framework that jointly optimizes task accuracy and calibration, alongside a novel evaluation benchmark supporting multi-reward calibration assessment. Experimental results demonstrate that the proposed approach significantly improves calibration performance and exhibits strong generalization across environments (e.g., local to web-based settings) and domains (e.g., mathematical reasoning).

This work addresses the lack of instance-level uncertainty modeling in machine learning predictions for online algorithm design. Methodologically, it is the first to systematically integrate probabilistic calibration—such as Platt scaling and isotonic regression—as a foundation for uncertainty quantification into classical online problems, including ski-rental and online job scheduling. It introduces a calibration-driven competitive ratio analysis framework that yields prediction-confidence-dependent theoretical guarantees. Theoretically, it establishes a quantitative relationship between calibration quality and competitive ratio performance, proving superiority over conventional uncertainty estimation—particularly under high-variance prediction regimes. Empirically, the proposed algorithms significantly outperform baselines on real-world job scheduling datasets and achieve optimal prediction-dependent performance in the ski-rental problem. Crucially, the theoretical guarantees align closely with empirical results, demonstrating both rigor and practical efficacy.

This study identifies a critical bidirectional risk of miscalibrated AI confidence in human-AI collaboration: overconfidence induces user misuse, while underconfidence triggers abandonment—both eroding trust, recommendation adherence, and decision effectiveness. Through a series of controlled behavioral experiments, integrating confidence calibration assessment with multidimensional trust and adoption metrics, we provide the first systematic empirical validation that confidence miscalibration significantly impairs collaborative quality. We further demonstrate that while transparency interventions improve users’ ability to detect miscalibration, they may paradoxically foster new forms of distrust. Consequently, we propose the design principle “calibration before transparency.” Results show that merely exposing uncalibrated confidence scores degrades trustworthy collaboration; only rigorously calibrated confidence estimates reliably support robust human-AI joint decision-making.

This study addresses the stagnation of enterprise AI initiatives in regulated financial institutions due to the absence of quantifiable evaluation criteria. Focusing on six document-intensive workflows, it systematically compares AI system performance across four model families and three tool configurations, distinguishing between demonstration and production environments. For the first time, it links deployment feasibility with human review rates. The authors propose a production-grade evaluation framework encompassing accuracy, reproducibility, traceability, and informative confidence, integrating multi-model comparison, confidence signals, source citation, and self-verification mechanisms. Experiments reveal that 56.1% of the 72 evaluated configurations meet production readiness thresholds. Incorporating source citation and confidence estimation reduces human review requirements to 49%, and adding self-verification further lowers this to 44%, albeit at the cost of reduced error tolerance.

This work addresses the problem of trust calibration in autonomous agents—specifically, how an agent should dynamically decide whether to act independently or seek human approval when using automated tools. The paper formalizes this challenge as a preference learning task for the first time. It introduces a policy gateway that maintains a Gaussian process posterior over the human’s risk tolerance function, employing a probit likelihood and an approximate Gaussian process classification model to infer preferences from binary approve/reject feedback. The agent actively queries human input at points of highest uncertainty, thereby establishing a three-region decision mechanism: “allow,” “block,” and “ask.” This approach extends the applicability of preference-based Bayesian optimization and achieves sample-efficient trust calibration, accurately partitioning the action space while substantially reducing unnecessary human interventions.