Action-conditional conformal prediction sets provide per-action safety guarantees for risk-averse policies that optimize conditional value-at-risk through pinball-loss minimization.
arXiv preprint arXiv:2409.00536 , year=
9 Pith papers cite this work. Polarity classification is still indexing.
9
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
Risk-controlled post-processing yields a threshold-structured policy that follows the baseline except where an oracle fallback sharply reduces conditional violation risk, achieving O(log n/n) expected excess risk in i.i.d. settings and exact risk control under exchangeability.
The work develops an iterative safe planner that adjusts conformal prediction bounds across policy updates via sensitivity analysis to maintain distribution-free safety guarantees despite interaction-induced distribution shifts.
Proposes imCP framework that uses independent samples from the invariant measure of a Markov process for conformal calibration in one-step and multi-step predictions of learned dynamical systems.
PAC-Bayesian bounds are derived for quadratic closed-loop control via SLS parameterization, yielding Chernoff certificates for posteriors over responses, a mean-response deployment result, and a data-driven learning algorithm.
State-dependent conformal prediction with genetic-algorithm state partitioning and branch-merging reachability produces tighter high-confidence perception-error bounds for scalable verification of neurally controlled autonomous systems.
OSCP optimizes conformal prediction score offsets via MILP minimization of an empirical region-size proxy for time-series, with validity guarantees and reduced computation versus prior methods.
Formal connections between PAC bounds for three data-driven reachability methods are established, with empirical results showing they are not interchangeable despite similarities.
citing papers explorer
-
Conformal Risk-Averse Decision Making with Action Conditional Guarantee
Action-conditional conformal prediction sets provide per-action safety guarantees for risk-averse policies that optimize conditional value-at-risk through pinball-loss minimization.
-
Risk-Controlled Post-Processing of Decision Policies
Risk-controlled post-processing yields a threshold-structured policy that follows the baseline except where an oracle fallback sharply reduces conditional violation risk, achieving O(log n/n) expected excess risk in i.i.d. settings and exact risk control under exchangeability.
-
Safe Planning in Interactive Environments via Iterative Policy Updates and Adversarially Robust Conformal Prediction
The work develops an iterative safe planner that adjusts conformal prediction bounds across policy updates via sensitivity analysis to maintain distribution-free safety guarantees despite interaction-induced distribution shifts.
-
Uncertainty Quantification via Invariant-Measure Conformal Prediction
Proposes imCP framework that uses independent samples from the invariant measure of a Markov process for conformal calibration in one-step and multi-step predictions of learned dynamical systems.
-
PAC-Bayesian Certificates for Quadratic Closed-Loop Control
PAC-Bayesian bounds are derived for quadratic closed-loop control via SLS parameterization, yielding Chernoff certificates for posteriors over responses, a mean-response deployment result, and a data-driven learning algorithm.
-
Statistical-Symbolic Verification of Perception-Based Autonomous Systems using State-Dependent Conformal Prediction
State-dependent conformal prediction with genetic-algorithm state partitioning and branch-merging reachability produces tighter high-confidence perception-error bounds for scalable verification of neurally controlled autonomous systems.
-
Efficient Quantification of Time-Series Prediction Error: Optimal Selection Conformal Prediction
OSCP optimizes conformal prediction score offsets via MILP minimization of an empirical region-size proxy for time-series, with validity guarantees and reduced computation versus prior methods.
-
Probably Approximately Correct (PAC) Guarantees for Data-Driven Reachability Analysis: A Theoretical and Empirical Comparison
Formal connections between PAC bounds for three data-driven reachability methods are established, with empirical results showing they are not interchangeable despite similarities.
- A PAC-Bayes Approach for Controlling Unknown Linear Discrete-time Systems