Tag
B-CARE is a prediction-blind framework that uses a fixed LLM verifier to assign classwise evidence scores for ranking potentially incorrect stance predictions, achieving a mean AUROC of 0.735 across 12 dataset-model settings and capturing 41% of errors within the top 20% highest-risk predictions.
该论文研究在已有输出侧信号的基础上，加入 LLM 隐藏状态对错误检测的增量预测价值，发现其在开放生成和多选任务上分别带来约 0.004 和 0.039 的 AUROC 提升，表明隐藏状态有额外信息但增益有限。
该论文提出两阶段诊断评估方法，检验以图结构表示的LLM推理链的结构可靠性，发现图的可靠性随拓扑、提示设计和模型族而变化，且图结构对不确定性估计的校准和节点级诊断有帮助，但作为提示上下文时收益有限。
提出Uncertainty-Aware Allocation (UAA)框架，通过兄弟节点语义分散度估计不确定性来动态分配LLM树搜索的每步扩展预算，在保持精度的同时将token使用量减少最多90%、延迟降低最多9倍，且可跨四种树搜索算法和开源/闭源模型泛化。
该论文研究不确定性度量何时能有效衡量 RAG 幻觉风险，发现当检索证据充分约束答案时，困惑度能达到 0.81–0.91 AUROC，而检索覆盖不足时同一指标会失效，提出不确定性是条件性的幻觉风险度量。
该论文提出 Control-TS，通过残差控制将预测分位数作为分布控制信号引入预训练扩散预测模型，并提出 CLPS 指标评估生成未来轨迹的下游预测效用，在八个基准上一致提升 CRPS 表现。
该论文首次在黑盒设定下研究 AMR 解析器输出的不确定性估计，利用 top-k beam-search 候选间的成对相似度评估可靠性，并通过 Prediction Rejection Ratio 分析解析性能与不确定性的关系，识别出性能与可靠性平衡最佳的解析器。
本文提出 RL-ARC，一个校准感知的训练框架，通过联合利用推理置信度与答案置信度来校准大型推理模型，缓解 RLVR 训练导致的过度自信问题，并在 ID 和 OOD 设置下提升校准表现而不显著牺牲推理性能。
提出了 ELF-BO 算法，利用目标函数评估的等待时间提前从超参数后验中采样，仅在观测到目标值后重新加权，从而以标准 BO 的决策延迟实现完全贝叶斯优化。在合成函数和真实应用中，ELF-BO 匹配完全贝叶斯方法的性能且延迟更低。
Researchers at Jefferson Lab and PNNL combine Gaussian process surrogate models with reinforcement learning to automate microwave frequency tuning for dynamically polarized targets, reporting nearly 2x improvement over expert operator actions.
This paper presents a predictive maintenance framework that uses deep learning and uncertainty estimation to improve remaining useful life estimation for semiconductor manufacturing, reducing maintenance costs.
A systematic study benchmarking training-free uncertainty quantification strategies for multimodal Large Language Models, categorizing methods into token-level, verbalized, and semantic approaches and finding optimal strategies depend on response length.
This paper evaluates entropy-based confidence signals for small language models, finding that semantic entropy provides a viable method to improve accuracy by up to 50 percentage points through selective routing to larger models.
This paper introduces an unsupervised metric based on attention dispersion to detect hallucinations in large language models, showing significant AUC improvements on mathematical reasoning benchmarks using the Qwen2.5 model family.
Introduces Kalman Delta Networks, which improve language modeling by reformulating linear attention as a linear-Gaussian state-space model with Kalman-filter updates to track memory uncertainty, yielding efficient approximations that outperform existing linear-attention models.
This paper proposes methods for detecting hallucinations in black-box LLMs by combining semantic entropy and token-level uncertainty signals, evaluating techniques like TopK, CoCoA, Gated, and Stacked across multiple benchmarks to find that no single method is universally strongest but Stacked often performs best.
This paper introduces Disease Continuum Positioning (DCP), a longitudinal Bayesian learning framework that continuously estimates Alzheimer's disease severity from neuroimaging data, providing an uncertainty-aware score for disease progression prediction.
This paper proposes a statistical model to efficiently estimate uncertainty dynamics in text generation, smoothing noisy resampling data to significantly reduce computational costs while maintaining accuracy in analyzing LLM reasoning chains.
This paper studies confidence estimation and selective prediction for financial named entity recognition under domain shift, evaluating BERT and LoRA-tuned Qwen models to enhance reliability across different input distributions like SEC filings and social media.
The paper proposes H2EDL, a hyper evidential deep learning model for hierarchical classification that captures uncertainty at multiple levels of a label hierarchy, reducing calibration error and improving the preservation of coarse categories in fine-grained recognition tasks.