TrustLoRA: Low-Rank Adaptation for Failure Detection under Out-of-distribution Data

Fei Zhu; Zhaoxiang Zhang

REVIEW 3 major objections 5 minor 45 references

TrustLoRA: Low-Rank Adaptation for Failure Detection under Out-of-distribution Data

T0 review · 3 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper claims that reliability knowledge for different failure types can be separated into low-rank adapters and then merged by simple weight addition, yielding a single model that rejects both misclassified covariate-shifted inputs…

desk verdict TrustLoRA is a genuine LoRA-arithmetic application to unified failure detection, but its own Table 6 undercuts the multi-task-learning advantage claim and the paper lacks error bars. read the letter →

arxiv 2504.14545 v1 pith:7EIBCABD submitted 2025-04-20 cs.LG

Fei Zhu , Zhaoxiang Zhang This is my paper · ORCID

classification cs.LG

keywords failuredetectionout-of-distributioncovariateshiftsemanticlow-rankadaptationLoRAarithmeticoutlierexposurereliableprediction

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

TrustLoRA argues that failure detection under distribution shift can be handled as an arithmetic operation on model weights rather than as one monolithic training problem. The paper trains two lightweight low-rank adapters on a frozen backbone, one capturing reliability under covariate shifts via AugMix and one capturing reliability under semantic shifts via outlier exposure, then adds their weight updates to produce a unified failure detector. The central claim is that this separated-then-merged procedure beats training a single model on both objectives, because the two objectives interfere when optimized jointly. A sympathetic reader would care because the approach promises a detector whose behavior can be tuned at inference time with one scalar, undone by subtraction, and extended to new failure sources by training only a small adapter.

What carries the argument

The load-bearing mechanism is the LoRA vector: the element-wise difference between the adapter weights before and after fine-tuning on a failure-specific objective (Eq. 5). Each vector records the direction the model moves to acquire one reliability skill. The framework adds the covariate-shift vector and the semantic-shift vector with weights $(1-\alpha)$ and $\alpha$, relying on the transfer of task arithmetic from full fine-tuning to low-rank adapters. A second mechanism is random-projection LoRA: the $A$ matrix is initialized once and frozen, only $B$ is trained, so trainable parameters stay below 1% of the backbone and the random seed is enough to reconstruct the projection.

What would settle it

Train the two LoRA adapters on a pair of objectives whose gradients are known to point in opposing directions for the same inputs, merge them with Eq. (6), and measure separate MisD and OOD-detection AUROC against a multi-task baseline. If the merged model is no better than multi-task training, or if the optimal $\alpha$ shifts dramatically across corruption types and datasets, the transfer of task arithmetic to LoRA is not general.

Watch

Extended reading notes

Core claim

The paper's central discovery is that reliability knowledge specific to one failure source can be isolated in a low-rank weight update and that these updates compose by simple addition. Given a frozen pretrained model, one LoRA module is fine-tuned with the AugMix objective (Eq. 3) to handle covariate shifts and another with outlier exposure (Eq. 4) to handle semantic shifts. The two trained LoRA vectors are then merged as $\tau = (1-\alpha)\tau_{\text{cov}} + \alpha \tau_{\text{sem}}$ (Eq. 6) and added to the backbone. The resulting TrustLoRA model rejects misclassified corrupted inputs and semantic OOD inputs with better AURC, FPR95, and AUC than single-objective tuning, post-hoc scores, and multi-task joint training across CIFAR-10/100, ImageNet-200/500, and ViT experiments. The scaling $\alpha$ acts as an inference-time dial between the two failure types, and subtracting a LoRA vector selectively removes one reliability skill.

Load-bearing premise

The framework assumes that adding the two independently trained low-rank weight updates produces a model that simultaneously keeps both reliability skills with no harmful mutual interference; if the objectives' updates cancel or conflict after linear combination, the claimed advantage over multi-task training collapses.

Editorial extensions

If this is right

A single merged model can reject both misclassified corrupted inputs and novel-class inputs, as measured across 90 mixtures of corruptions and semantic OOD sets.
The coefficient $\alpha$ in Eq. (6) lets a deployment choose the balance between covariate-shift rejection and semantic-shift rejection at inference time without retraining.
Negating a LoRA vector removes one reliability skill (e.g., OOD detection) while leaving misclassification detection largely intact.
New failure sources can be added by fine-tuning one small adapter and merging, rather than retraining or fully fine-tuning the model, with trainable parameters under 1% of the backbone.
The improvement persists on large-scale ImageNet and ViT backbones and is robust to the LoRA rank and to the choice of auxiliary outlier dataset.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A natural extension the paper leaves implicit is incremental deployment: as new failure sources appear over time, each could be captured by its own LoRA and merged cumulatively, with rollback possible by subtraction if a new source conflicts with old ones.
The arithmetic view predicts that the optimal $\alpha$ is not a global constant but depends on the corruption type and dataset, because the geometric relation between the two LoRA vectors changes; the paper's fixed $\alpha = 0.5$ may hide this variability.
Because only the $B$ matrix is trained under a fixed random projection, the merged LoRA is a linear function of the $B$ matrices, so the combined model's behavior on any input is a convex interpolation of the two adapters' behaviors; this could be characterized analytically and might fail when the two skills require opposing feature directions.
The paper does not explore score functions beyond MSP; applying energy or nearest-neighbor scores to the features of the merged model could further improve separation of correct and incorrect covariate-shifted samples, since the adapters reshape features, not just confidence scores.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Dear [Colleague],

TrustLoRA is a genuine combination that deserves a serious look: it trains separate LoRA adapters for covariate-shift robustness (AugMix) and semantic-shift rejection (outlier exposure), then merges them through task arithmetic, with a random-projection variant that only trains B. That is a real application of LoRA and task arithmetic to failure detection, not a new framework, but it is a useful one. The experiments are broad—CIFAR-10/100, ImageNet, ViT, many baselines—and the flexibility analysis (alpha sweep, negation editing, rank/auxiliary-data robustness) is a strength. The paper earns credit for that.

The soft spots are real, though not fatal. Most important: the paper's own Table 6 undercuts its claim that LoRA arithmetic outperforms multi-task learning. On CIFAR-100 severity-1, TrustLoRA has AURC 129.64 vs 138.92 for AugMix+OE (Full FT); on severity-2, TrustLoRA is 149.14 vs 142.66—worse. Since the paper calls AURC the most important metric, this inconsistency matters, and severity-3 is omitted entirely. No error bars or seed-level variance anywhere, so even the severity-1 edge could be noise. The F-AUC metric definition has a typo (AUCcov repeated) and is ambiguous about what it actually measures. The theoretical motivation for LoRA arithmetic is thin: it leans on prior task-arithmetic work without analyzing when linear merging of two LoRA adapters fails, especially under interference from ranks, random projections, or stronger/weaker auxiliary OOD sets. No code is released, which compounds the reproducibility concern.

That said, the central idea—separating reliability knowledge into low-rank adapters and consolidating it via arithmetic—remains plausible. The issue is not that it cannot work; it is that the current evidence is inconsistent and under-reported. A serious referee should ask for a reconciled comparison on all severities, confidence intervals, a clarified F-AUC, and ideally code.

Who is this for? Anyone working on failure detection, OOD detection, or parameter-efficient fine-tuning. It is a useful citation for the LoRA-arithmetic application. I would send it to review; it deserves referee time, but conditional on those fixes. My vote would be borderline accept with major revision.

Best, [You]

Referee Report

3 major / 5 minor

Summary. The paper proposes TrustLoRA, a framework for unified failure detection under both covariate and semantic shifts. Two low-rank adapters are trained separately, one with an AugMix objective for covariate robustness and one with an outlier-exposure objective for semantic OOD detection, and then merged via LoRA arithmetic (Eq. 6) with a mixing weight alpha. The authors argue that separating and later consolidating failure-specific reliability knowledge yields better unified failure detection than training a single model jointly on both objectives, while also providing flexibility, parameter efficiency, and recoverability. Experiments cover CIFAR-10/100-C with six semantic OOD sets, ImageNet-200/500, and ViT-B16, together with ablations on rank, random projection, auxiliary data, and LoRA negation.

Significance. If the central claim holds, TrustLoRA is a practically useful contribution: it offers a parameter-efficient way to combine reliability knowledge from different failure sources, with an explicit interpolation parameter alpha to control the trade-off at inference time, and it provides a first demonstration of low-rank adapters for separating and consolidating reliability knowledge. The paper's strengths are its broad empirical coverage, the comparison against many training-time and post-hoc baselines, and the ablations on rank, random projection, and auxiliary outlier sets. The main weakness is that the evidence for the load-bearing advantage over multi-task learning is internally inconsistent and lacks uncertainty quantification, and the theoretical justification for LoRA arithmetic is informal.

major comments (3)

[Section 5.1, Table 6] The claim that 'our LoRA arithmetic outperforms the multi-task learning, i.e., AugMix+OE (Full FT) in Table 6' is not supported by the reported AURC values on the paper's primary metric. On CIFAR-100 severity-2, TrustLoRA has AURC 149.14, which is worse than AugMix+OE (Full FT) at 142.66, and severity-3 results are not reported for these baselines. Since Section 5.1 calls AURC 'the most important metric for failure detection evaluation,' the single table supporting the separation-versus-joint-training advantage is internally inconsistent across severities. Please report the full severity range and seed-level variance (or confidence intervals) so the reader can assess whether the severity-1 advantage is robust.
[Section 4.2, Eq. (6)] The key assumption that LoRA arithmetic transfers task arithmetic is not justified. Equation (6) adds two LoRA vectors that are trained independently with different random projection matrices unless the same A initialization is explicitly enforced, yet there is no analysis or experiment showing that linear merging preserves both reliability skills under this subspace mismatch. The paper should either provide a derivation or explicit conditions under which Eq. (6) holds, or add ablations probing seed sensitivity, ranks, and auxiliary outlier sets of varying strength. Without this, the claimed advantage over multi-task learning rests on an untested assumption.
[Section 5, Metrics] The evaluation mixture used for AURC, FPR95, AUC, and F-AUC is underspecified. The text says that equal numbers of misclassified covariate-shifted data and semantic OOD data are kept, but it does not state how many correctly classified covariate-shifted samples are included, whether the equality is imposed before or after conditioning on misclassification, or how the results over the 90 mixtures are aggregated. Since the unified failure detection metrics depend on the relative proportions of correct covariate-shifted, misclassified covariate-shifted, and semantic OOD samples, the results in Tables 1, 2, and 4 cannot be reproduced without this information.

minor comments (5)

[Section 5, Metrics] The definition of F-AUC contains a typo: 'AUCcov denotes the AUC value of separating correct and incorrect covariate-shifted data and AUCcov denotes the AUC value of separating covariate-shifted and semantic-shifted data' should use AUCsem for the second component, matching the formula that follows.
[Table 6 caption] The caption says 'Comparison with more baselines and multi-task learning on CIFAR-100, severity-1,' but the table also reports Severity-2 columns; the caption should describe both severities.
[Section 5.1, ImageNet paragraph] The text says 'Results in Table 5 suggest that our method yields strong failure detection performance,' but the corresponding table is numbered Table 3 in the manuscript.
[Section 4.2, Eq. (5)] The notation {θ_pre, θ_LoRA,t} in Eq. (5) is ambiguous; if it denotes concatenation or vectorization of the pre-trained and LoRA parameters, that should be defined explicitly.
[Throughout] There are several typos: 'ARUC' should be 'AURC' in Section 5.1, 'shits' should be 'shifts' in Section 5.1, 'the enhance' should be 'to enhance' in Related Work, and 'which others' should be 'while others' in Related Work.

Circularity Check

0 steps flagged · score 1.0 of 10

No material circularity: LoRA vectors are trained with independent objectives and merged by standard task arithmetic; no fitted parameter is relabeled as a prediction.

full rationale

TrustLoRA's derivation chain is self-contained and does not reduce to its inputs. The two LoRA vectors are trained with established, external objectives: AugMix (Eq. 3) for covariate-shift reliability and outlier exposure (Eq. 4) for semantic-shift reliability, and the paper explicitly states 'we do not propose novel failure-specific learning objectives in this paper.' The consolidation step (Eq. 6) is the standard task-arithmetic linear combination introduced by Ilharco et al. [34], applied here to LoRA parameters; it is not derived from, nor fitted to, the failure-detection metric being reported. The hyperparameters alpha=0.5 and rank r=4 are fixed and stated, not optimized on the test set. Evaluation is against external benchmarks (CIFAR-10/100-C, ImageNet, standard semantic OOD sets), and the central claim that LoRA arithmetic outperforms multi-task learning is an empirical comparison in Table 6, not a consequence of the equations. The paper's self-citations to prior failure-detection work (RCL, OpenMix, confidence-estimation work) are used for context and baselines rather than as the load-bearing justification for the new merging mechanism. The internal inconsistency in Table 6 on CIFAR-100 severity-2 and the absence of confidence intervals are correctness and robustness concerns, not circularity. No step in the claimed derivation is equivalent to its own input by construction.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the transferability of task arithmetic to LoRA adapters plus the validity of the two building-block losses (AugMix and OE). These are imported from prior work rather than derived here, and the linear merge assumption is untested outside the reported settings.

free parameters (3)

Mixing weight alpha = 0.5
Scales the contribution of the semantic-shift LoRA vector in Eq. (6); set to 0.5 universally after showing robustness across [0.4, 0.6] in Fig. 5.
LoRA rank r = 4
Rank of the low-rank adapters; set to 4 for all main experiments, with robustness shown across r=2,4,8 in Table 7.
Loss weight lambda in AugMix and OE objectives = not reported
Appears in Eqs. (3) and (4); the value is not stated in the paper, presumably taken from the original AugMix and OE implementations.

assumptions (4)

domain assumption AugMix with cross-entropy and JS consistency improves robustness to corruption without hurting accuracy.
Used in Eq. (3) to train the covariate LoRA; taken from prior work [32].
domain assumption Outlier exposure with uniform KL loss teaches rejection of semantic OOD samples.
Used in Eq. (4) to train the semantic LoRA; taken from prior work [6].
domain assumption LoRA weight vectors are linearly composable for skill transfer.
Central mechanism in Eq. (6); assumed from task arithmetic literature [34] without new proof.
domain assumption Covariate and semantic shift input spaces are sufficiently separable for OOD detection to be possible.
Problem formulation in Section 3 relies on this separability to make rejection well-defined.

how reviews work

0 comments

Cite this review

Pith. "Pith review of TrustLoRA: Low-Rank Adaptation for Failure Detection under Out-of-distribution Data." pith.science (2026). https://pith.science/paper/7EIBCABD

@misc{pith2026250414545,
  author       = {Pith},
  title        = {Pith review of: TrustLoRA: Low-Rank Adaptation for Failure Detection under Out-of-distribution Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7EIBCABD}},
  note         = {Machine review of arXiv:2504.14545}
}

read the original abstract

Reliable prediction is an essential requirement for deep neural models that are deployed in open environments, where both covariate and semantic out-of-distribution (OOD) data arise naturally. In practice, to make safe decisions, a reliable model should accept correctly recognized inputs while rejecting both those misclassified covariate-shifted and semantic-shifted examples. Besides, considering the potential existing trade-off between rejecting different failure cases, more convenient, controllable, and flexible failure detection approaches are needed. To meet the above requirements, we propose a simple failure detection framework to unify and facilitate classification with rejection under both covariate and semantic shifts. Our key insight is that by separating and consolidating failure-specific reliability knowledge with low-rank adapters and then integrating them, we can enhance the failure detection ability effectively and flexibly. Extensive experiments demonstrate the superiority of our framework.

Figures

Figures reproduced from arXiv: 2504.14545 by the authors.

**Figure 2.** Covariate shifts complicate failure detection. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** Illustration of the proposed reliability arithmetic framework. (Left) We freeze the pre [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: Change of rejection ability when fine-tuning the [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

**Figure 5.** Figure 5: Flexibility of controlling the strength of reliability edition on CIFAR-100. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

**Figure 6.** Figure 6: Accurate forgetting of OOD detection ability [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

**Figure 7.** Figure 7: MisD ability on clean ID. ular OOD datasets with in total of 2715 OOD samples; for covariate shifts, we use the corruption type Frost with severity-1. Results in [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]

**Figure 8.** Figure 8: Risk-coverage curves on the mixture of ID, covari [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 42 canonical work pages

[1]

Zhang, J

Y. Zhang, J. Hu, D. Wen, W. Deng, Unsupervised evaluation for out-of-distribution detection, Pattern Recognition 160 (2025) 111212. 1

work page 2025
[2]

Zhang, Y

Y. Zhang, Y. Chen, C. Fang, Q. Wang, J. Wu, J. Xin, Learning from open-set noisy labels based on multi-prototype modeling, Pattern Recognition 157 (2025) 110902. 1

work page 2025
[3]

F. Zhu, S. Ma, Z. Cheng, X.-Y. Zhang, Z. Zhang, C.-L. Liu, Open-world machine learning: A review and new outlooks, arXiv preprint arXiv:2403.01759 (2024). 1

arXiv 2024
[4]

H. Bai, G. Canal, X. Du, J. Kwon, R. D. Nowak, Y. Li, Feed two birds with one scone: Exploiting wild data for both out-of-distribution generalization and detection, in: International Conference on Machine Learning, PMLR, 2023, pp. 1454–1471. 1, 3

work page 2023
[5]

Hendrycks, T

D. Hendrycks, T. Dietterich, Benchmarking neural network robustness to common corruptions and perturbations, in: International Conference on Learning Representations, 2018. 1, 7

work page 2018
[6]

Hendrycks, M

D. Hendrycks, M. Mazeika, T. Dietterich, Deep anomaly detection with outlier exposure, in: International Conference on Learning Representations, 2018. 1, 3, 6, 8

work page 2018
[7]

Y. Wang, K. Yu, G. Xiang, F. Cao, J. Liang, Discovering causally invariant features for out-of- distribution generalization, Pattern Recognition 150 (2024) 110338. 1

work page 2024
[8]

W. Liu, H. Yu, Y. Wang, B. Liu, D. Tao, H. Chen, Iw-vit: Independence-driven weighting vision transformer for out-of-distribution generalization, Pattern Recognition 161 (2025) 111308. 1

work page 2025

Show all 45 references

[9]

S. Ma, F. Zhu, Z. Cheng, X.-Y. Zhang, Towards trustworthy dataset distillation, Pattern Recog- nition 157 (2025) 110875. 1, 3

2025
[10]

J. Wei, G. Wang, S. Zhang, Fine-grained medical image out-of-distribution detection through multi-view feature uncertainty and adversarial sample generation, Pattern Recognition (2025) 111401. 1, 3

2025
[11]

Y. Jia, J. Li, G. Zhao, S. Liu, W. Sun, L. Lin, G. Li, Enhancing out-of-distribution detection via diversified multi-prototype contrastive learning, Pattern Recognition 161 (2025) 111214. 1, 3

2025
[12]

Cheng, F

Z. Cheng, F. Zhu, X.-Y. Zhang, C.-L. Liu, Average of pruning: Improving performance and stability of out-of-distribution detection, arXiv preprint arXiv:2303.01201 (2023). 1

2023 arXiv
[13]

F. Zhu, Z. Cheng, X.-Y. Zhang, C.-L. Liu, Z. Zhang, Rcl: Reliable continual learning for unified failure detection, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 12140–12150. 2, 3, 8

2024
[14]

Zhu, X.-Y

F. Zhu, X.-Y. Zhang, R.-Q. Wang, C.-L. Liu, Learning by seeing more classes, IEEE Transactions on Pattern Analysis and Machine Intelligence 45 (6) (2022) 7477–7493. 2

2022
[15]

F. Zhu, Z. Cheng, X.-Y. Zhang, C.-L. Liu, Openmix: Exploring outlier samples for misclassifica- tion detection, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 12074–12083. 2, 3, 8

2023
[16]

Y. Li, Y. Chen, X. Yu, D. Chen, X. Shen, Sure: Survey recipes for building reliable and robust deep networks, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 17500–17510. 2, 3, 8 14

2024
[17]

W. Liu, X. Wang, J. Owens, Y. Li, Energy-based out-of-distribution detection, Advances in neural information processing systems 33 (2020) 21464–21475. 2, 8

2020
[18]

Y. Sun, Y. Ming, X. Zhu, Y. Li, Out-of-distribution detection with deep nearest neighbors, in: International Conference on Machine Learning, PMLR, 2022, pp. 20827–20840. 2, 8

2022
[19]

J. Park, Y. G. Jung, A. B. J. Teoh, Nearest neighbor guidance for out-of-distribution detection, in: Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 1686–1695. 2, 8

2023
[20]

J.-H. Kim, S. Yun, H. O. Song, Neural relation graph: A unified framework for identifying label noise and outlier data, Advances in Neural Information Processing Systems (2023). 2, 8

2023
[21]

X. Liu, Y. Lochman, C. Zach, Gen: Pushing the limits of softmax-based out-of-distribution detection, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 23946–23955. 2, 8

2023
[22]

Djurisic, N

A. Djurisic, N. Bozanic, A. Ashok, R. Liu, Extremely simple activation shaping for out-of- distribution detection, in: The Eleventh International Conference on Learning Representations,
[23]

Cheng, F

Z. Cheng, F. Zhu, X.-Y. Zhang, C.-L. Liu, Breaking the limits of reliable prediction via generated data, International Journal of Computer Vision (2024) 1–27. 3

2024
[24]

F. Zeng, Z. Cheng, F. Zhu, H. Wei, X.-Y. Zhang, Local-prompt: Extensible local prompts for few-shot out-of-distribution detection, in: The Thirteenth International Conference on Learning Representations, 2025. 3

2025
[25]

F. Zeng, Z. Cheng, F. Zhu, X.-Y. Zhang, Towards efficient and general-purpose few-shot misclassification detection for vision-language models, arXiv preprint arXiv:2503.20492 (2025). 3

2025 arXiv
[26]

J. Cen, D. Luan, S. Zhang, Y. Pei, Y. Zhang, D. Zhao, S. Shen, Q. Chen, The devil is in the wrongly- classified samples: Towards unified open-set recognition, arXiv preprint arXiv:2302.04002 (2023). 3, 8

2023 arXiv
[27]

K. He, X. Zhang, S. Ren, J. Sun, Deep residual learning for image recognition, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778. 4

2016
[28]

P . F. Jaeger, C. T. Lüth, L. Klein, T. J. Bungert, A call to reflect on evaluation practices for failure detection in image classification, in: The Eleventh International Conference on Learning Representations, 2022. 4, 9

2022
[29]

J. Kim, J. Koo, S. Hwang, A unified benchmark for the unknown detection capability of deep neural networks, Expert Systems with Applications 229 (2023) 120461. 4

2023
[30]

Gueta, E

A. Gueta, E. Venezian, C. Raffel, N. Slonim, Y. Katz, L. Choshen, Knowledge is a region in weight space for fine-tuned language models, in: Findings of the Association for Computational Linguistics: EMNLP 2023, 2023, pp. 1350–1370. 4

2023
[31]

E. J. Hu, P . Wallis, Z. Allen-Zhu, Y. Li, S. Wang, L. Wang, W. Chen, et al., Lora: Low-rank adaptation of large language models, in: International Conference on Learning Representations,
[32]

Hendrycks, N

D. Hendrycks, N. Mu, E. D. Cubuk, B. Zoph, J. Gilmer, B. Lakshminarayanan, Augmix: A simple data processing method to improve robustness and uncertainty, in: International Conference on Learning Representations, 2019. 5, 8

2019
[33]

Dasgupta, A

S. Dasgupta, A. Gupta, An elementary proof of a theorem of johnson and lindenstrauss, Random Structures & Algorithms 22 (1) (2003) 60–65. 6 15

2003
[34]

Ilharco, M

G. Ilharco, M. T. Ribeiro, M. Wortsman, L. Schmidt, H. Hajishirzi, A. Farhadi, Editing models with task arithmetic, in: The Eleventh International Conference on Learning Representations,
[35]

Geifman, R

Y. Geifman, R. El-Yaniv, Selective classification for deep neural networks, in: Advances in Neural Information Processing Systems, 2017, pp. 4878–4887. 7

2017
[36]

Hendrycks, K

D. Hendrycks, K. Gimpel, A baseline for detecting misclassified and out-of-distribution ex- amples in neural networks, in: International Conference on Learning Representations, 2016. 7

2016
[37]

Pinto, H

F. Pinto, H. Yang, S.-N. Lim, P . Torr, P . K. Dokania, Regmixup: Mixup as a regularizer can surprisingly improve accuracy and out distribution robustness, in: Advances in Neural Infor- mation Processing Systems, 2022. 7, 9

2022
[38]

J. Moon, J. Kim, Y. Shin, S. Hwang, Confidence-aware learning for deep neural networks, in: international conference on machine learning, PMLR, 2020, pp. 7034–7044. 7, 9

2020
[39]

H. Wei, R. Xie, H. Cheng, L. Feng, B. An, Y. Li, Mitigating neural network overconfidence with logit normalization, in: International conference on machine learning, PMLR, 2022, pp. 23631–23644. 7, 9

2022
[40]

Hendrycks, S

D. Hendrycks, S. Basart, M. Mazeika, A. Zou, J. Kwon, M. Mostajabi, J. Steinhardt, D. Song, Scaling out-of-distribution detection for real-world settings, in: International Conference on Machine Learning, 2022. 8

2022
[41]

Bitterwolf, M

J. Bitterwolf, M. Müller, M. Hein, In or out? fixing imagenet out-of-distribution detection evaluation, in: Proceedings of the 40th International Conference on Machine Learning, 2023, pp. 2471–2506. 9

2023
[42]

Zhu, X.-Y

F. Zhu, X.-Y. Zhang, Z. Cheng, C.-L. Liu, Revisiting confidence estimation: Towards reliable failure prediction, IEEE Transactions on Pattern Analysis and Machine Intelligence (2023). 10, 11

2023
[43]

Dosovitskiy, L

A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, et al., An image is worth 16x16 words: Transformers for image recognition at scale, in: International Conference on Learning Representations, 2020. 10

2020
[44]

Xia, C.-S

G. Xia, C.-S. Bouganis, Augmenting the softmax with additional confidence scores for improved selective classification with out-of-distribution data, International Journal of Computer Vision (2024) 1–39. 11

2024
[45]

Zhang, J

J. Zhang, J. Yang, P . Wang, H. Wang, Y. Lin, H. Zhang, Y. Sun, X. Du, K. Zhou, W. Zhang, et al., Openood v1. 5: Enhanced benchmark for out-of-distribution detection, arXiv preprint arXiv:2306.09301 (2023). 12 16

2023 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

Zhang, J

Y. Zhang, J. Hu, D. Wen, W. Deng, Unsupervised evaluation for out-of-distribution detection, Pattern Recognition 160 (2025) 111212. 1

work page 2025

[2] [2]

Zhang, Y

Y. Zhang, Y. Chen, C. Fang, Q. Wang, J. Wu, J. Xin, Learning from open-set noisy labels based on multi-prototype modeling, Pattern Recognition 157 (2025) 110902. 1

work page 2025

[3] [3]

F. Zhu, S. Ma, Z. Cheng, X.-Y. Zhang, Z. Zhang, C.-L. Liu, Open-world machine learning: A review and new outlooks, arXiv preprint arXiv:2403.01759 (2024). 1

arXiv 2024

[4] [4]

H. Bai, G. Canal, X. Du, J. Kwon, R. D. Nowak, Y. Li, Feed two birds with one scone: Exploiting wild data for both out-of-distribution generalization and detection, in: International Conference on Machine Learning, PMLR, 2023, pp. 1454–1471. 1, 3

work page 2023

[5] [5]

Hendrycks, T

D. Hendrycks, T. Dietterich, Benchmarking neural network robustness to common corruptions and perturbations, in: International Conference on Learning Representations, 2018. 1, 7

work page 2018

[6] [6]

Hendrycks, M

D. Hendrycks, M. Mazeika, T. Dietterich, Deep anomaly detection with outlier exposure, in: International Conference on Learning Representations, 2018. 1, 3, 6, 8

work page 2018

[7] [7]

Y. Wang, K. Yu, G. Xiang, F. Cao, J. Liang, Discovering causally invariant features for out-of- distribution generalization, Pattern Recognition 150 (2024) 110338. 1

work page 2024

[8] [8]

W. Liu, H. Yu, Y. Wang, B. Liu, D. Tao, H. Chen, Iw-vit: Independence-driven weighting vision transformer for out-of-distribution generalization, Pattern Recognition 161 (2025) 111308. 1

work page 2025