The minimax rate of estimating second-order calibration error is Õ(1/√n) with a matching Ω(1/√n) lower bound, enabled by analyticity from the sech kernel and yielding the first finite-sample guarantee for second-order Platt scaling.
hub
International conference on machine learning , pages=
23 Pith papers cite this work. Polarity classification is still indexing.
23
Pith papers citing it
hub tools
citation-role summary
method 2
citation-polarity summary
years
2026 23roles
method 2polarities
use method 2representative citing papers
ProcCtrlBench introduces an ontology of 11 defect types across 4 categories plus control preservation metrics to evaluate LLM coding agent trajectories on 200 cases from AndroidBench, TerminalBench, and SWE-bench-Verified.
VHYDRO is a support-safe variational hybrid filter that jointly recovers continuous latent states, discrete contact modes, and sparse port-Hamiltonian laws per regime while preventing loss of feasible transitions.
AstroAlertBench evaluates multimodal LLMs on astronomical classification accuracy, reasoning, and honesty using real ZTF alerts, revealing that high accuracy often diverges from self-assessed reasoning quality.
mPL measures attacker-aligned privacy leakage from joint data releases and AmPL provides an adaptive way to bound it with low utility cost in ML settings.
Trajectory geometry of MLP write-vectors, summarized by 11 scale-invariant features in a sparse linear probe, beats MSP for selective abstention in LLM multiple-choice QA.
A 149M-parameter distributional energy-based verifier with low-rank adapter ensemble reduces constraint violations in structured LLM reasoning and outperforms or matches much larger models on five benchmarks.
Conflicting biomedical evidence triggers order-dependent prediction flips in RAG LLMs, and a new abstention score combining confidence with conflict detection raises selective accuracy by 7-33 points in the hardest conditions.
SynIB is an information-theoretic objective that adds a penalty for unimodal confidence to standard task loss, improving accuracy on synergy-dependent examples by up to 7.8% across synthetic XOR tasks and five real-world multimodal benchmarks.
RisCoSet applies multiple hypothesis testing to construct risk-controlling partial-program prediction sets for LLM code generation, achieving up to 24.5% less code removal than prior methods at equivalent risk levels.
A fair conformal classification method guarantees conditional coverage on adaptively identified subgroups defined via learned representations.
ReSIDe generalizes logit-based confidence scores to intermediate layers of synthetic image detectors and uses preference optimization to aggregate them, cutting area under the risk-coverage curve by up to 69.55% under covariate shifts.
Safety certification of dynamical systems is reformulated as direct classification via kernel embeddings on trajectories, bypassing recursive DP to avoid error compounding and support non-Markovian dynamics.
LaaB improves LLM hallucination detection by mapping self-judgment labels back into neural feature space and using mutual learning under logical consistency constraints between responses and meta-judgments.
Perturb-and-Correct generates epistemically diverse predictors from a single pretrained network via hidden-layer perturbations followed by affine least-squares corrections that enforce agreement on calibration data.
Conditional risk calibration reduces to standard regression and is distinct from probability calibration.
Align-Cultura introduces the CULTURAX dataset and shows that culturally fine-tuned LLMs improve joint HHH scores by 4-6%, cut cultural failures by 18%, and gain 10-12% efficiency with minimal leakage.
Social identity markers in medical questions degrade LLM accuracy and uncertainty calibration, producing a calibration crisis that is non-additive for intersectional cases.
Avoiding CenterLoss improves OOD detection via multi-scale Mahalanobis on L2-normalized features, yielding 0.9483 AUROC on CIFAR-10 while preserving competitive in-distribution accuracy.
C3RL is a new RL algorithm combining correctness, calibration, and reference accuracy rewards to improve LLM confidence calibration, enabling CAS to outperform majority voting with up to 12.33x lower inference cost.
Annotation disagreement on toxic language can be moderately predicted from textual features, with high-opposition items proving harder for models to estimate accurately.
ClaHF converts instance labels into preference signals via candidate predictions and a reward model, then applies RL optimization to improve text classification accuracy and calibration.
citing papers explorer
-
The Minimax Rate of Second-Order Calibration
The minimax rate of estimating second-order calibration error is Õ(1/√n) with a matching Ω(1/√n) lower bound, enabled by analyticity from the sech kernel and yielding the first finite-sample guarantee for second-order Platt scaling.
-
ProcCtrlBench: Evaluating Process-Level Defects and Control Preservation in LLM Coding Agents
ProcCtrlBench introduces an ontology of 11 defect types across 4 categories plus control preservation metrics to evaluate LLM coding agent trajectories on 200 cases from AndroidBench, TerminalBench, and SWE-bench-Verified.
-
Support-Safe Variational Hybrid Filtering for Contact-Mode and Sparse-Law Recovery
VHYDRO is a support-safe variational hybrid filter that jointly recovers continuous latent states, discrete contact modes, and sparse port-Hamiltonian laws per regime while preventing loss of feasible transitions.
-
AstroAlertBench: Evaluating the Accuracy, Reasoning, and Honesty of Multimodal LLMs in Astronomical Classification
AstroAlertBench evaluates multimodal LLMs on astronomical classification accuracy, reasoning, and honesty using real ZTF alerts, revealing that high accuracy often diverges from self-assessed reasoning quality.
-
Metric-Normalized Posterior Leakage (mPL): Attacker-Aligned Privacy for Joint Consumption
mPL measures attacker-aligned privacy leakage from joint data releases and AmPL provides an adaptive way to bound it with low utility cost in ML settings.
-
Reading Calibrated Uncertainty from Language Model Trajectories
Trajectory geometry of MLP write-vectors, summarized by 11 scale-invariant features in a sparse linear probe, beats MSP for selective abstention in LLM multiple-choice QA.
-
Distributional Energy-Based Models for Uncertainty-Aware Structured LLM Reasoning
A 149M-parameter distributional energy-based verifier with low-rank adapter ensemble reduces constraint violations in structured LLM reasoning and outperforms or matches much larger models on five benchmarks.
-
When Evidence Conflicts: Uncertainty and Order Effects in Retrieval-Augmented Biomedical Question Answering
Conflicting biomedical evidence triggers order-dependent prediction flips in RAG LLMs, and a new abstention score combining confidence with conflict detection raises selective accuracy by 7-33 points in the hardest conditions.
-
SynIB: Informational Bottleneck for Maximizing Synergy in Multimodal Learning
SynIB is an information-theoretic objective that adds a penalty for unimodal confidence to standard task loss, improving accuracy on synergy-dependent examples by up to 7.8% across synthetic XOR tasks and five real-world multimodal benchmarks.
-
Uncertainty Quantification for LLM-based Code Generation
RisCoSet applies multiple hypothesis testing to construct risk-controlling partial-program prediction sets for LLM code generation, achieving up to 24.5% less code removal than prior methods at equivalent risk levels.
-
Fair Conformal Classification via Learning Representation-Based Groups
A fair conformal classification method guarantees conditional coverage on adaptively identified subgroups defined via learned representations.
-
Post-hoc Selective Classification for Reliable Synthetic Image Detection
ReSIDe generalizes logit-based confidence scores to intermediate layers of synthetic image detectors and uses preference optimization to aggregate them, cutting area under the risk-coverage curve by up to 69.55% under covariate shifts.
-
Safety Certification is Classification
Safety certification of dynamical systems is reformulated as direct classification via kernel embeddings on trajectories, bypassing recursive DP to avoid error compounding and support non-Markovian dynamics.
-
Logical Consistency as a Bridge: Improving LLM Hallucination Detection via Label Constraint Modeling between Responses and Self-Judgments
LaaB improves LLM hallucination detection by mapping self-judgment labels back into neural feature space and using mutual learning under logical consistency constraints between responses and meta-judgments.
-
Perturb and Correct: Post-Hoc Ensembles using Affine Redundancy
Perturb-and-Correct generates epistemically diverse predictors from a single pretrained network via hidden-layer perturbations followed by affine least-squares corrections that enforce agreement on calibration data.
-
Calibrating conditional risk
Conditional risk calibration reduces to standard regression and is distinct from probability calibration.
-
AlignCultura: Towards Culturally Aligned Large Language Models?
Align-Cultura introduces the CULTURAX dataset and shows that culturally fine-tuned LLMs improve joint HHH scores by 4-6%, cut cultural failures by 18%, and gain 10-12% efficiency with minimal leakage.
-
Calibrated? Not for Everyone: How Sexual Orientation and Religious Markers Distort LLM Accuracy and Confidence in Medical QA
Social identity markers in medical questions degrade LLM accuracy and uncertainty calibration, producing a calibration crisis that is non-additive for intersectional cases.
-
Don't Collapse Your Features: Why CenterLoss Hurts OOD Detection and Multi-Scale Mahalanobis Wins
Avoiding CenterLoss improves OOD detection via multi-scale Mahalanobis on L2-normalized features, yielding 0.9483 AUROC on CIFAR-10 while preserving competitive in-distribution accuracy.
-
Scaling with Confidence: Calibrating Confidence of LLMs for Adaptive Test Time Scaling
C3RL is a new RL algorithm combining correctness, calibration, and reference accuracy rewards to improve LLM confidence calibration, enabling CAS to outperform majority voting with up to 12.33x lower inference cost.
-
Quantifying and Predicting Disagreement in Graded Human Ratings
Annotation disagreement on toxic language can be moderately predicted from textual features, with high-opposition items proving harder for models to estimate accurately.
-
ClaHF: A Human Feedback-inspired Reinforcement Learning Framework for Improving Classification Tasks
ClaHF converts instance labels into preference signals via candidate predictions and a reward model, then applies RL optimization to improve text classification accuracy and calibration.
- Instance-Adaptive Online Multicalibration