{"total":12,"items":[{"citing_arxiv_id":"2606.29952","ref_index":12,"ref_count":1,"confidence":0.98,"is_internal_anchor":true,"paper_title":"Exploiting Local Flatness for Efficient Out-of-Distribution Detection","primary_cat":"cs.LG","submitted_at":"2026-06-29T08:27:27+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Fold is a post-hoc OOD detector that exploits larger feature-Hessian curvature on OOD inputs together with partial feature normalization and a self-supervised AutoFold calibration scheme.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.06156","ref_index":3,"ref_count":1,"confidence":0.98,"is_internal_anchor":true,"paper_title":"Trust-Aware Predictive Emissions Monitoring for Gas Turbine Fleets with Limited Labelled Data","primary_cat":"cs.LG","submitted_at":"2026-06-04T13:28:58+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":4.0,"formal_verification":"none","one_line_summary":"A multi-head RNN framework with learned confidence, ensemble uncertainty, auxiliary predictions, distance analysis, and diagnostics produces calibrated trust scores for NOx prediction, reducing MAE from 0.202 to 0.070 on the top 10% confidence subset.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2606.00069","ref_index":22,"ref_count":1,"confidence":0.98,"is_internal_anchor":true,"paper_title":"Invascal: Inverse-Vacuity Self-Calibration for Uncertainty-Aware LiDAR Range-View Semantic Segmentation","primary_cat":"cs.RO","submitted_at":"2026-05-20T16:39:29+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Introduces an architecture-agnostic Adapter Head and Invascal self-calibration objective to produce calibrated evidential uncertainty estimates for LiDAR range-view semantic segmentation while preserving accuracy.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.01063","ref_index":20,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"GEODE: Angle-Adaptive OOD Detection with Universal Scorer Compatibility","primary_cat":"cs.LG","submitted_at":"2026-05-01T19:56:29+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"GEODE uses per-sample cosine-similarity scaling in a norm loss to preserve feature geometry for universal scorer-compatible OOD detection, matching or exceeding OE performance on CIFAR benchmarks.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2605.00640","ref_index":51,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Knowing when to trust machine-learned interatomic potentials","primary_cat":"cs.LG","submitted_at":"2026-05-01T13:21:56+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"PROBE recasts MLIP uncertainty quantification as selective classification by training a compact discriminative classifier on frozen per-atom backbone embeddings, yielding a reliability probability that tracks actual error better than ensemble disagreement.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.21546","ref_index":3,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Component-Based Out-of-Distribution Detection","primary_cat":"cs.CV","submitted_at":"2026-04-23T11:19:39+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"CoOD decomposes inputs into components and applies Component Shift Score plus Compositional Consistency Score to improve detection of both standard and compositional out-of-distribution data.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"grained OOD shifts that occur beyond the focused regions. As will be analyzed theoretically in this section, CoOD can address these issues from three aspects: (1) explicitly in- cludes diverse component representations to mitigate the loss of detection-relevant local information (Section 3.2.1), (2) reduces confusion from irrelevant context and noise (Section 3.2.2), and (3) supplies compositional evidence of inter-component relationships (Biederman, 1987). 3.2.1. INTRODUCINGCOMPONENTREPRESENTATIONS Prior work (Hu et al., 2024) shows that auxiliary information can improve OOD detection. Following widely used setup (Jiang et al., 2024; Chen et al., 2024; Le et al., 2025), we define anexistence scorefor each component p as syp ="},{"citing_arxiv_id":"2604.13262","ref_index":2,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"Rethinking Uncertainty in Segmentation: From Estimation to Decision","primary_cat":"cs.CV","submitted_at":"2026-04-14T19:52:05+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"Uncertainty optimization alone misses most safety gains; a decision-stage deferral policy removes up to 80% segmentation errors at 25% pixel deferral with cross-dataset robustness, while calibration does not improve decision quality.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.04614","ref_index":34,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"A Clinical Point Cloud Paradigm for In-Hospital Mortality Prediction from Multi-Level Incomplete Multimodal EHRs","primary_cat":"cs.LG","submitted_at":"2026-04-06T12:03:36+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":7.0,"formal_verification":"none","one_line_summary":"HealthPoint represents clinical events as points in a 4D space (content, time, modality, case) and applies low-rank relational attention to achieve state-of-the-art mortality prediction from multi-level incomplete multimodal EHRs.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2604.01798","ref_index":36,"ref_count":1,"confidence":0.9,"is_internal_anchor":false,"paper_title":"A deep learning pipeline for PAM50 subtype classification using histopathology images and multi-objective patch selection","primary_cat":"cs.CV","submitted_at":"2026-04-02T09:13:24+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"An optimization-based deep learning pipeline selects informative patches from H&E whole-slide images to classify breast cancer into PAM50 subtypes, achieving F1 scores of 0.88 internally and 0.80 externally.","context_count":1,"top_context_role":"background","top_context_polarity":"background","context_text":"uncertainty can better manage diagnostic risk, allow for selective expert review, and support safer and more transparent clinical deployment in pathology. 4 Although uncertainty estimation has not been explicitly addressed in previous PAM50 classification studies [10-16], it has been recognized as both clinically and technically important in the broader medical AI literature. The generative segmentation approach proposed by Kohl et al. [36] addresses intrinsic image ambiguity through the modeling of probable outputs; however, its segmentation -specific design and complexity hinder its direct application to classification tasks. To identify predictions that are inaccurate or out of distributio n, Hendrycks and Gimpel [37] employed confidence -based uncertainty estimation; nevertheless, their heuristic technique lacks adequate calibration."},{"citing_arxiv_id":"2511.19996","ref_index":6,"ref_count":1,"confidence":0.98,"is_internal_anchor":true,"paper_title":"RankOOD -- Class Ranking-based Out-of-Distribution Detection","primary_cat":"cs.LG","submitted_at":"2025-11-25T07:02:56+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"RankOOD detects out-of-distribution samples by training a model to predict fixed class-specific ranking permutations via the Plackett-Luce loss, achieving a 4.3% FPR95 reduction on near-OOD TinyImageNet.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2511.11934","ref_index":2,"ref_count":1,"confidence":0.98,"is_internal_anchor":true,"paper_title":"A Systematic Analysis of Out-of-Distribution Detection Under Representation and Training Paradigm Shifts","primary_cat":"cs.LG","submitted_at":"2025-11-14T23:18:13+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":5.0,"formal_verification":"none","one_line_summary":"Benchmark across architectures and shift regimes finds OOD detector rankings shift with representation collapse; proposes NC-based shortlist predictor and PCA filter without extra OOD data.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null},{"citing_arxiv_id":"2504.14129","ref_index":37,"ref_count":1,"confidence":0.98,"is_internal_anchor":true,"paper_title":"PVLM: Parsing-Aware Vision Language Model with Dynamic Contrastive Learning for Zero-Shot Deepfake Attribution","primary_cat":"cs.CV","submitted_at":"2025-04-19T01:11:46+00:00","verdict":"UNVERDICTED","verdict_confidence":"LOW","novelty_score":6.0,"formal_verification":"none","one_line_summary":"PVLM combines parsing-aware vision-language modeling with dynamic contrastive learning to enable fine-grained zero-shot attribution of deepfakes to unseen generators and outperforms prior methods on a new benchmark.","context_count":0,"top_context_role":null,"top_context_polarity":null,"context_text":null}],"limit":50,"offset":0}