arXiv:2606. 05551v1 Announce Type: cross Abstract: Reliable decision making pipelines powered by machine learning models require uncertainty quantification (UQ) methods that come with explicit safety guarantees.
arXiv:2607. 02206v1 Announce Type: cross Abstract: Predictions are increasingly used to guide high-stakes decisions, from treatment selection to policy making.
arXiv:2510. 07750v3 Announce Type: replace-cross Abstract: Robust optimization safeguards decisions against uncertainty by optimizing against worst-case scenarios, yet their effectiveness hinges on a prespecified robustness level that is often chosen ad hoc, leading to either insufficient protection or overly conservative and costly solutions.
The paper introduces RATTL (Risk-Adversarial Total-Reward Learning), a framework that adjusts an agent’s caution based on epistemic uncertainty by using a Bayesian posterior over dynamics and a Wasserstein ambiguity set whose radius depends on that posterior. As evidence accumulates, the radius shrinks, smoothly transitioning the agent’s behavior from worst-case robustness to risk-neutral reward maximization. The authors prove a Safety Sandwich theorem showing RATTL’s value lies between the uninformed robust value and the full-knowledge optimum, and demonstrate the method on a binary-hazard example where the criterion reduces to Conditional Value-at-Risk.
arXiv:2603. 02491v3 Announce Type: replace-cross Abstract: As artificial agents become increasingly capable, what internal structure is necessary for an agent to act competently under uncertainty?
The paper investigates risk‑averse decision making where an agent chooses actions under uncertainty about the system state, using optimized certainty equivalent (OCE) metrics that encompass mean‑variance risk and CVaR. For known distributions, the optimal policy simplifies to a prediction‑set‑based solution for CVaR, linking it to conformal prediction sets. When distributions are unknown, the authors propose a data‑driven calibration method that employs a synthetic likelihood model and held‑out data to achieve high‑probability OCE risk control, and they demonstrate the method on two wireless beamforming scenarios.
arXiv:2606. 00320v1 Announce Type: new Abstract: We present an online, distribution-free framework for controlling the Conditional Value-at-Risk (CVaR), extending conformal tail risk control to non-stationary and adversarial environments.
arXiv:2604. 26836v3 Announce Type: replace Abstract: Predictive safety filters (PSFs) leverage model predictive control to enforce constraint satisfaction during deep reinforcement learning (RL) exploration, yet their reliance on first-principles models or Gaussian processes limits scalability and broader applicability.
The paper introduces a score‑calibrated robustness framework that transforms any fixed point predictor into a decision‑relevant uncertainty representation using distribution‑free conformal calibration. By employing the conformal score as the core unit of robustness, the authors derive both reliability‑based robust optimization and target‑oriented Conformal Robust Satisficing formulations, linking them through a shared robust decision frontier and a fragility measure. Experiments on synthetic data and a real online‑grocery inventory case study demonstrate the framework’s ability to improve reliability, reduce costs, and provide interpretable uncertainty scales for black‑box predictors.
arXiv:2507. 20068v2 Announce Type: replace Abstract: Off-policy evaluation (OPE) methods estimate the value of a new reinforcement learning (RL) policy prior to deployment.
arXiv:2607. 05620v1 Announce Type: cross Abstract: In many decision-making settings, new interventions are acceptable only if they do not reduce outcomes below some established threshold.
arXiv:2603. 02196v3 Announce Type: replace Abstract: An agent must try new behaviors to explore and improve.
The paper introduces a new approach to safety in contextual bandits with continuous actions by enforcing high‑probability constraints on the realized cost rather than on its expectation. It proposes the High‑Probability Constrained UCB algorithm, which balances optimistic reward exploration with pessimistic safety estimation. The authors provide theoretical regret guarantees for linear models and extend the analysis to general function classes, demonstrating experimentally that realized‑cost constraints significantly reduce safety violations compared to expected‑cost baselines.