Pith. sign in

REVIEW 3 major objections 6 minor 49 references

FEVER-OOD: Free Energy Vulnerability Elimination for Robust Out-of-Distribution Detection

T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

Pith's one-line read Free-energy out-of-distribution detection has a provable blind spot, and this paper shows how to shrink it.

desk verdict Null-space vulnerability is real and worth remembering; the least-singular-value derivation solves the wrong objective, but the empirical study is solid enough to warrant serious review. read the letter →

arxiv 2412.01596 v1 pith:OGZB5EBQ submitted 2024-12-02 cs.CV

classification cs.CV
keywords out-of-distributiondetectionfreeenergyscorenullspacevulnerabilityleastsingularvaluefeatureregularizationenergy-basedOODImageNet-100object-level
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

The paper argues that the free energy score, a widely used uncertainty measure for out-of-distribution (OOD) detection, contains an inherent blind spot: when a classifier's last layer is linear, certain feature-space differences between in-distribution and OOD samples are invisible to the score, and other differences produce only a minimal energy change. The root cause is algebraic, not empirical: any displacement lying in the null space of the last-layer weight matrix leaves the logits and hence the free energy exactly unchanged, and the smallest non-zero change is controlled by the least singular value of that matrix. To counter this, the paper inserts a dimension-reducing layer before the classifier head and adds regularizers that raise the least singular value and balance the condition number. Empirically, these interventions improve OOD detection across several energy-based baselines and achieve a new reported best on ImageNet-100, where the false positive rate at 95% true positive rate drops from 39.98% to 35.93%. The significance is practical: models deployed in open-set settings can now be made less prone to silently accepting outlier inputs.

What carries the argument

The central object is the linear last layer W_cls ∈ $R^{{d'×K}}$ of the classifier, because the free energy score depends on features only through the logits W_cls^T h(x). The rank-nullity theorem supplies the blind subspace: if d' > K, then Null(W_cls^T) is non-trivial, and any displacement δ in that subspace gives W_cls^T δ = 0 and hence an identical free energy. The singular value decomposition supplies the weakest detection direction: over all unit displacements orthogonal to the null space, the minimum logit change is d_b σ_min(W_cls), so outliers near the least singular vector are the hardest to separate. The paper's countermeasure combines three mechanisms: r-NSR, a learned linear map g': $R^{{d'}}$ → R^r with r < d' that shrinks the null space before the head; the least singular value regularizer, which adds λ_LSV σ_min(W_cls)^{-1} to the training loss; and the condition number regularizer, which adds λ_CN κ(W_cls) to balance energy changes over all directions.

What would settle it

Take a trained classifier with a linear last layer, extract real feature vectors, fix a displacement norm d_b, and compare the actual free energy change along the least singular vector with the change along the next few singular vectors; if the least singular vector no longer gives the smallest free energy change, the proxy argument behind Eq. (12) fails. More decisively, run a feature-space search for a displacement with norm at least d_b whose free energy change is smaller than d_b σ_min(W_cls); finding even one such displacement in a FEVER-OOD-trained network would break the claimed guarantee.

Watch

Extended reading notes

Core claim

The paper establishes that, for a classifier whose final layer is linear, f(x) = W_cls^T h(x), the free energy F(x) depends on the feature vector h(x) only through the logits. Consequently, any OOD feature displacement δ with h(v) = h(x) + δ that lies in the null space of W_cls^T leaves the logits, and therefore the free energy, exactly unchanged, so distinct feature representations can be OOD-invisible; this is the null space vulnerability (NSV). For displacements orthogonal to the null space, the smallest possible change in the logits, and by the paper's proxy argument the smallest change in free energy, is d_b σ_min(W_cls), attained along the direction of the least singular vector; this is the least singular value vulnerability (LSVV). The paper's proposed remedy, FEVER-OOD, inserts an extra linear layer that projects features down to dimension r < d' before the head (r-NSR), adds a least singular value regularizer that maximizes σ_min(W_cls), and optionally a condition number regularizer that equalizes energy changes across directions. Across CIFAR-10, CIFAR-100, ImageNet-100, and object-detection benchmarks built on VOS, FFS, and Dream-OOD baselines, the combined method consistently improves FPR95 and AUROC, with the headline ImageNet-100 result reaching 35.93% average FPR95 versus the 39.98% baseline.

Load-bearing premise

The argument depends on the free energy's log-sum-exp nonlinearity being monotone enough that the direction of minimal logit change is also the direction of minimal free energy change, and on the last layer being full rank; either failing could leave an OOD direction that is still nearly invisible.

Editorial extensions

If this is right

  • Any energy-based OOD detector built on a classifier whose feature dimension exceeds its class count inherits a non-trivial null space and therefore a set of OOD displacements that are invisible to the free energy score.
  • Reducing the feature dimension before the final linear layer shrinks that blind subspace and improves OOD detection without degrading in-distribution accuracy, provided the reduction is not extreme.
  • Raising the least singular value of the last layer increases the free energy change along the previously weakest direction, so OOD samples near that direction become detectable.
  • On ImageNet-100, applying FEVER-OOD to the Dream-OOD baseline yields 35.93% average FPR95 versus 39.98% and 93.12 AUROC versus 91.64, which the paper reports as state of the art for that benchmark.
  • The same null-space and least-singular-value reasoning applies to object-level OOD detection, where the paper reports improved FPR95 and AUROC on MS-COCO and OpenImages with PASCAL VOC as in-distribution data.

Reading between the lines

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

  • Beyond the paper's direct experiments, the null-space argument transfers to any confidence score that depends on features through a linear head, so the dimension-reduction and singular-value regularizers could plausibly strengthen softmax-based and logit-based OOD detectors as well.
  • The least singular value analysis also describes a potential attack surface: an adversary who can manipulate feature representations could place OOD samples along the least singular vector to make them look in-distribution, and the LSV regularizer effectively raises the cost of that attack.
  • A testable extension would be to freeze the feature extractor and apply only the null-space-reducing layer and least-singular-value regularizer at the head, isolating how much of the benefit comes from the head geometry rather than from joint retraining.
  • The ablation results suggest the method's benefit is largest when the reduction target r is chosen just below the feature dimension; extremely small r (for example r = 10 on CIFAR-10) degrades in-distribution accuracy, indicating that the null-space cure can be overdosed.
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.

Referee Report

3 major / 6 minor

Summary. The paper studies out-of-distribution (OOD) detection based on the free energy score. It claims two inherent vulnerabilities of free-energy scoring for classifiers with a linear last layer: a null-space vulnerability (NSV), where feature-space displacements in the null space of the last-layer weight matrix leave the free energy exactly unchanged, and a least-singular-value vulnerability (LSVV), where the minimal free-energy change for a displacement of norm at least d_b is claimed to be d_b \sigma_{\min}(W_{\text{cls}}). To mitigate these, it proposes r-null-space reduction (r-NSR), a least-singular-value regularizer (LSVR), and a condition-number regularizer (CNR). The method is evaluated on VOS, FFS, and Dream-OOD baselines for CIFAR-10, CIFAR-100, ImageNet-100, and PASCAL VOC object detection, reporting improved FPR95 and AUROC, with a headline ImageNet-100 result of 35.93% FPR95 versus 39.98% for the Dream-OOD baseline.

Significance. If the theoretical claims were correct, the paper would identify a previously unexplored blind spot of free-energy OOD detection and offer a principled fix. The NSV observation itself is correct and is a useful, simple insight: for a linear last layer with d' > K, any feature-space displacement in the null space of W_{\text{cls}}^T indeed leaves the free energy unchanged. The paper also provides a broad empirical study across several baselines, datasets, and tasks, with code, ablations, and deterministic training. However, the central LSVV derivation is mathematically incorrect: Eq. (12) solves a proxy problem about logit changes, not the stated free-energy minimization, and the two objectives are not equivalent. In addition, hyperparameters are selected post hoc on the same OOD test sets and results are reported from single runs without error bars, so the empirical claims are weaker than stated. The paper overclaims by calling the result state-of-the-art without comparison to recent OOD methods beyond the three chosen baselines.

major comments (3)
  1. [§4.2, Eqs. (10)–(12)] Equation (12) does not solve the minimization problem stated in Eq. (10). The free energy F(x) = -log \sum_k \exp(f_k(x)) is unchanged whenever \sum_k \exp(f_k + r_k) = \sum_k \exp(f_k). For K \ge 2, there exist nonzero vectors r of arbitrarily large norm satisfying this identity; for example, when f(x)=0 one may take r_1 = t and r_2 = \log(2 - e^t) with t < \log 2, so the sum of exponentials remains 2. Since W_{\text{cls}}^T restricted to Null(W_{\text{cls}}^T)^\perp is a bijection onto \mathbb{R}^K whenever W_{\text{cls}} has rank K, any such logit perturbation r is realized by some δ in the feasible set of Eq. (10), and by taking t sufficiently negative one obtains \|δ\| \ge d_b. Hence the minimum in Eq. (10) is 0, not d_b \sigma_{\min}(W_{\text{cls}}), for any full-rank W_{\text{cls}} with K \ge 2. The quantity in Eq. (12) bounds the smallest possible change in the logits, which is neither necessary nor sufficient for a small free-energy change. Consequently, the claim in §4.3 that maximizing \sigma_{\min} “guarantee[s] detectable energy variations” is unsupported. The LSVV section must be rewritten, either by downgrading the claim to a statement about logit-norm change or by deriving a genuine bound on the free-energy difference; in its current form the central theoretical contribution is not valid.
  2. [§5.3 and §6, Tables 1–3] The reported gains are based on post hoc selection of hyperparameters on the very OOD test sets used for evaluation. Section 6 states that “The reported values for λLSV and λCN are for the best obtained models,” and Section 5.3 evaluates a grid of values and selects configurations. No held-out OOD validation split is used, and all results come from a single deterministic run, so no uncertainty or significance information is available. The headline improvement of 35.93% versus 39.98% FPR95 is therefore a selected maximum over the grid rather than a pre-registered or cross-validated comparison. The authors should either report results on an OOD validation set separated from the test set, or provide multiple seeds with error bars and a clear statement of the selection procedure.
  3. [§6, ImageNet-100 and CIFAR-100 paragraphs] The empirical evidence that the LSVR/CNR regularizers are the source of improvement is weak and at times contradicted by the paper’s own tables. In Table 3, Dream-OOD-100-NSR without any regularizer attains an average FPR95 of 35.60%, while the claimed best model with LSVR attains 35.93% (AUROC improves from 92.64 to 93.12 but FPR95 worsens). In Table 2, adding LSVR to Dream-OOD-100-NSR degrades FPR95 from 42.77% to 66.15%. The paper’s explanation that pretraining is responsible is not tested. Given that the theoretical justification for the regularizers is in question, the contribution of LSVR/CNR relative to r-NSR alone needs to be established with controlled comparisons; as it stands, the results largely support the conclusion that r-NSR is the main driver of the reported gains.
minor comments (6)
  1. [Abstract and Table 3] The abstract reports an average OOD false positive rate of 35.83%, while Table 3 and the conclusion report 35.93%; the numbers should be reconciled.
  2. [Table 1] The OOD dataset heading “SHVN” is a typo for “SVHN”.
  3. [§5.3 and Table 3] The text states that for ImageNet-100 the authors use λLSV = 0.001 and λCN = 0.01 based on CIFAR-100 experiments, but Table 3 reports λLSV = 0.01 and λCN = 0.001 for the best configurations; this mismatch should be corrected.
  4. [§6 and Conclusion] The “state-of-the-art” claim is not supported by a comparison with recent OOD detection methods beyond VOS, FFS, and Dream-OOD; the claim should be qualified or the comparison expanded.
  5. [Supplementary Material] The supplementary material contains several typos, including “regualrizing”, “failrs”, and “fos”; these should be fixed in a revision.
  6. [Figure 2] The label “Safe Direction” in Figure 2 is not defined in the text or caption; the figure should clarify what makes a direction safe.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the vulnerability analysis is self-contained linear algebra, Eq. (11) is explicitly a proxy, and the only self-citation is non-load-bearing.

full rationale

The paper's core derivation (Secs. 4.1-4.3) does not fit any parameter and then rename it as a prediction. NSV follows directly from W_cls^T δ = 0, which forces identical logits and hence identical free energy. LSVV is obtained by minimizing ∥W_cls^T δ∥ over δ with ∥δ∥ ≥ d_b and δ orthogonal to the null space; the solution d_b σ_min(W_cls) is a standard singular-value identity, and the paper explicitly labels Eq. (11) a 'proxy minimum' rather than claiming it is equivalent to Eq. (10). No load-bearing step is justified by a self-citation: reference [16], which shares authors, is cited only as an existing energy-based OOD method alongside VOS, FFS and Dream-OOD, and does not support the uniqueness or validity of the proposed regularizers. The empirical sections compare against external baselines and report ablations, so the improvement claims are independent of the theoretical derivation. The mathematical gap noted in the skeptic analysis (that Eq. (12) bounds logit-norm change, whereas the free energy can be constant on level surfaces in logit space even for δ with W_cls^T δ ≠ 0) is a correctness risk in the theory, not a circularity: it would weaken the LSVV claim, but it does not make the claimed result an input of the derivation. Likewise, the choice of λ_LSV, λ_CN and r on the evaluated OOD sets is a benchmark-selection concern, not a circular derivation. Accordingly, no circular step is identified.

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

The central claims rest on a linear last layer, a distance-based notion of OOD, a proxy connecting free energy to logit norm, and a full rank assumption. The only fitted quantities are the regularizer weights and the NSR dimension, selected post hoc on the test sets.

free parameters (3)
  • Least singular value regularizer weight λ_LSV = 0.001 to 1.0 (best per model/dataset)
    Controls the penalty on σ_min(W_cls)^{-1}; the reported best values are selected from ablations on the same OOD test sets (Tables 1-3, Supp. Figs. 9-13).
  • Condition number regularizer weight λ_CN = 0.001 to 1.0 (best per model/dataset)
    Controls the penalty on the condition number κ(W_cls); values are selected post hoc on the test benchmarks, with some instability reported for large λ_CN.
  • NSR output dimension r = 10, 32, 64, 96, 100, 114, 128, 256, 512, 768 depending on dataset
    The reduced feature dimension is a free hyperparameter chosen by evaluating several values on the same test OOD sets (Tables 1-4).
assumptions (4)
  • domain assumption The final layer of the classifier is linear: f(x) = W_cls^T h(x) (Eq. 5).
    The null space and singular value derivations rely on this form; most vision classifiers use a linear classification head.
  • domain assumption OOD samples are far from in-distribution features, so h(v) = h(x) + δ with ||δ|| ≥ db (Eqs. 6-7).
    This justifies considering non-zero differences δ and defines the threat model for the vulnerability analysis.
  • ad hoc to paper The change in free energy is well approximated by the change in logit norm, reducing Eq. (10) to min ||W_cls^T δ|| (Eq. 11).
    The paper calls this a proxy minimum; the log-sum-exp nonlinearity is not rigorously accounted for before deriving Eq. (12).
  • domain assumption W_cls has full column rank, so nullity = d' - K and a square r = K layer has zero nullity.
    Used to claim that 100-NSR eliminates the null space for ImageNet-100; a rank-deficient square matrix would leave a residual null space.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FEVER-OOD: Free Energy Vulnerability Elimination for Robust Out-of-Distribution Detection." pith.science (2026). https://pith.science/paper/OGZB5EBQ

@misc{pith2026241201596,
  author       = {Pith},
  title        = {Pith review of: FEVER-OOD: Free Energy Vulnerability Elimination for Robust Out-of-Distribution Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OGZB5EBQ}},
  note         = {Machine review of arXiv:2412.01596}
}
read the original abstract

Modern machine learning models, that excel on computer vision tasks such as classification and object detection, are often overconfident in their predictions for Out-of-Distribution (OOD) examples, resulting in unpredictable behaviour for open-set environments. Recent works have demonstrated that the free energy score is an effective measure of uncertainty for OOD detection given its close relationship to the data distribution. However, despite free energy-based methods representing a significant empirical advance in OOD detection, our theoretical analysis reveals previously unexplored and inherent vulnerabilities within the free energy score formulation such that in-distribution and OOD instances can have distinct feature representations yet identical free energy scores. This phenomenon occurs when the vector direction representing the feature space difference between the in-distribution and OOD sample lies within the null space of the last layer of a neural-based classifier. To mitigate these issues, we explore lower-dimensional feature spaces to reduce the null space footprint and introduce novel regularisation to maximize the least singular value of the final linear layer, hence enhancing inter-sample free energy separation. We refer to these techniques as Free Energy Vulnerability Elimination for Robust Out-of-Distribution Detection (FEVER-OOD). Our experiments show that FEVER-OOD techniques achieve state of the art OOD detection in Imagenet-100, with average OOD false positive rate (at 95% true positive rate) of 35.83% when used with the baseline Dream-OOD model.

Figures

Figures reproduced from arXiv: 2412.01596 by the authors.

Figure 1
Figure 1. FEVER-OOD (right, green) improves baseline Free Energy-based OOD detection methods (left, blue). OOD is essential in developing deployable machine learn￾ing systems, to ensure valid inference only on data drawn from same distribution used in training [22, 33, 39], real￾world deployments often involve scenarios where models encounter unseen OOD classes that could otherwise under￾mine decision stability and hence wide… view at source ↗
Figure 2
Figure 2. Vulnerabilities of Free Energy-based OOD detection. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Null space reduction. We add an extra layer [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: MS-COCO objects detected on OOD images by VOS baseline [ [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Feature Space UMAP Projection for models trained on CIFAR-10. Top row corresponds to the VOS [ [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]
Figure 6
Figure 6. Figure 6: Feature Space t-SNE Projection for models trained on CIFAR-10. Top row corresponds to the VOS [ [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: Free energy change by its distance to the centroid of the feature vectors of an in-distribution category along different directions. [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Free energy change for varying the contribution of the Null Space (NS) component and the Null Space Perpendicular (NSP) [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: Ablations of the loss weight for the LSV and CN regularizers in FEVER-OOD for VOS, using CIFAR-10 as in-distribution (ID). [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]
Figure 10
Figure 10. Figure 10: Ablations of the loss weight for the LSV and CN regularizers in FEVER-OOD for FFS, using CIFAR-10 as in-distribution (ID). [PITH_FULL_IMAGE:figures/full_fig_p015_10.png]
Figure 11
Figure 11. Figure 11: Ablations of the loss weight for the LSV and CN regularizers in FEVER-OOD for VOS, using CIFAR-100 as in-distribution [PITH_FULL_IMAGE:figures/full_fig_p016_11.png]
Figure 12
Figure 12. Figure 12: Ablations of the loss weight for the LSV and CN regularizers in FEVER-OOD for FFS, using CIFAR-100 as in-distribution [PITH_FULL_IMAGE:figures/full_fig_p016_12.png]
Figure 13
Figure 13. Figure 13: Ablations of the loss weight for the LSV and CN regularizers in FEVER-OOD for Dream-OOD, using CIFAR-100 as in [PITH_FULL_IMAGE:figures/full_fig_p016_13.png]
Figure 14
Figure 14. Figure 14: Additional visualization of detected objects on the OOD images (from OpenImages [ [PITH_FULL_IMAGE:figures/full_fig_p017_14.png]
Figure 15
Figure 15. Figure 15: Additional visualization of detected objects on the OOD images (from MS-COCO [ [PITH_FULL_IMAGE:figures/full_fig_p018_15.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 44 canonical work pages

  1. [1]

    Feed two birds with one scone: Exploiting wild data for both out-of-distribution gen- eralization and detection

    Haoyue Bai, Gregory Canal, Xuefeng Du, Jeongyeol Kwon, Robert D Nowak, and Yixuan Li. Feed two birds with one scone: Exploiting wild data for both out-of-distribution gen- eralization and detection. In International Conference on Machine Learning, pages 1454–1471. PMLR, 2023. 2

  2. [2]

    AHA: Human-Assisted Out-of-Distribution Generalization and Detection

    Haoyue Bai, Jifan Zhang, and Robert Nowak. Aha: Human-assisted out-of-distribution generalization and detec- tion. arXiv preprint arXiv:2410.08000, 2024. 2

  3. [3]

    Describing textures in the wild

    Mircea Cimpoi, Subhransu Maji, Iasonas Kokkinos, Sammy Mohamed, and Andrea Vedaldi. Describing textures in the wild. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition , pages 3606–3613,

  4. [4]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl- vain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition at scale. In International Conference on Learning Representa- tions, 2021. 3

  5. [5]

    Unknown-aware object detection: Learning what you don’t know from videos in the wild

    Xuefeng Du, Xin Wang, Gabriel Gozum, and Yixuan Li. Unknown-aware object detection: Learning what you don’t know from videos in the wild. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13678–13688, 2022. 1

  6. [6]

    V os: Learning what you don’t know by virtual outlier synthesis

    Xuefeng Du, Zhaoning Wang, Mu Cai, and Yixuan Li. V os: Learning what you don’t know by virtual outlier synthesis. In International Conference on Learning Representations ,

  7. [7]

    Dream the impossible: Outlier imagination with diffusion mod- els

    Xuefeng Du, Yiyou Sun, Xiaojin Zhu, and Yixuan Li. Dream the impossible: Outlier imagination with diffusion mod- els. In Advances in Neural Information Processing Systems,

  8. [8]

    How does unlabeled data provably help out-of-distribution detection? arXiv preprint arXiv:2402.03502, 2024

    Xuefeng Du, Zhen Fang, Ilias Diakonikolas, and Yixuan Li. How does unlabeled data provably help out-of-distribution detection? arXiv preprint arXiv:2402.03502, 2024. 1

Show all 49 references
  1. [9]

    Mark Everingham, Luc Van Gool, Christopher K. I. Williams, John M. Winn, and Andrew Zisserman. The pascal visual object classes (voc) challenge. International Journal of Computer Vision, 88(2):303–338, 2010. 5, 8, 12, 17, 18

  2. [10]

    Rich feature hierarchies for accurate object detection and semantic segmentation

    Ross Girshick, Jeff Donahue, Trevor Darrell, and Jitendra Malik. Rich feature hierarchies for accurate object detection and semantic segmentation. In Proceedings of the IEEE con- ference on computer vision and pattern recognition , pages 580–587, 2014. 5

  3. [11]

    Deep Learning

    Ian Goodfellow, Yoshua Bengio, and Aaron Courville. Deep Learning . MIT Press, 2016. http : / / www . deeplearningbook.org. 3

  4. [12]

    Waslander

    Ali Harakeh and Steven L. Waslander. Estimating and evalu- ating regression predictive uncertainty in deep object detec- tors. In International Conference on Learning Representa- tions, 2021. 5

  5. [13]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 770–778, 2016. 3, 5, 12

  6. [14]

    A baseline for detect- ing misclassified and out-of-distribution examples in neural networks

    Dan Hendrycks and Kevin Gimpel. A baseline for detect- ing misclassified and out-of-distribution examples in neural networks. In International Conference on Learning Repre- sentations, 2022. 2

  7. [15]

    Generalized odin: Detecting out-of-distribution image with- out learning from out-of-distribution data

    Yen-Chang Hsu, Yilin Shen, Hongxia Jin, and Zsolt Kira. Generalized odin: Detecting out-of-distribution image with- out learning from out-of-distribution data. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10951–10960, 2020. 1

  8. [16]

    Isaac-Medina, Y .F.A

    B.K.S. Isaac-Medina, Y .F.A. Gaus, N. Bhowmik, and T.P. Breckon. Towards open-world object-based anomaly detec- tion via self-supervised outlier synthesis. In Proc. European Conference on Computer Vision. Springer, 2024. 1, 2, 3, 4

  9. [17]

    Training ood detectors in their natural habitats

    Julian Katz-Samuels, Julia B Nakhleh, Robert Nowak, and Yixuan Li. Training ood detectors in their natural habitats. In International Conference on Machine Learning , pages 10848–10865. PMLR, 2022. 2

  10. [18]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky. Learning multiple layers of features from tiny images. Technical report, University of Toronto, 2009. 5, 12

  11. [19]

    Normalizing flow based feature synthe- sis for outlier-aware object detection

    Nishant Kumar, Sini ˇsa ˇSegvi´c, Abouzar Eslami, and Ste- fan Gumhold. Normalizing flow based feature synthe- sis for outlier-aware object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5156–5165, 2023. 1, 2, 3, 4, 5, 8,...

  12. [20]

    The open images dataset v4: Unified image classification, object detection, and visual relationship detection at scale

    Alina Kuznetsova, Hassan Rom, Neil Alldrin, Jasper Ui- jlings, Ivan Krasin, Jordi Pont-Tuset, Shahab Kamali, Stefan Popov, Matteo Malloci, Alexander Kolesnikov, et al. The open images dataset v4: Unified image classification, object detection, and visual relationship detection...

  13. [21]

    Simple and scalable predictive uncertainty estima- tion using deep ensembles

    Balaji Lakshminarayanan, Alexander Pritzel, and Charles Blundell. Simple and scalable predictive uncertainty estima- tion using deep ensembles. Advances in Neural Information Processing Systems, 30, 2017. 1

  14. [22]

    A survey on out-of-distribution detection in NLP

    Hao Lang, Yinhe Zheng, Yixuan Li, Jian SUN, Fei Huang, and Yongbin Li. A survey on out-of-distribution detection in NLP. Transactions on Machine Learning Research, 2024. 1

  15. [23]

    A simple unified framework for detecting out-of-distribution samples and adversarial attacks

    Kimin Lee, Kibok Lee, Honglak Lee, and Jinwoo Shin. A simple unified framework for detecting out-of-distribution samples and adversarial attacks. Advances in Neural Infor- mation Processing Systems, 31, 2018. 1, 2

  16. [24]

    Enhancing the re- liability of out-of-distribution image detection in neural net- works

    Shiyu Liang, Yixuan Li, and R Srikant. Enhancing the re- liability of out-of-distribution image detection in neural net- works. In 6th International Conference on Learning Repre- sentations, 2018, 2018. 1, 2

  17. [25]

    Microsoft coco: Common objects in context

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Doll´ar, and C Lawrence Zitnick. Microsoft coco: Common objects in context. In Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceeding...

  18. [26]

    Energy-based out-of-distribution detection

    Weitang Liu, Xiaoyun Wang, John Owens, and Yixuan Li. Energy-based out-of-distribution detection. Advances in 9 Neural Information Processing Systems , 33:21464–21475,

  19. [27]

    Yuval Netzer, Tao Wang, Adam Coates, Alessandro Bis- sacco, Bo Wu, and Andrew Y . Ng. Reading digits in natural images with unsupervised feature learning. In NIPS Work- shop on Deep Learning and Unsupervised Feature Learning 2011, 2011. 5

  20. [28]

    Deep neural networks are easily fooled: High confidence predictions for unrecognizable images

    Anh Nguyen, Jason Yosinski, and Jeff Clune. Deep neural networks are easily fooled: High confidence predictions for unrecognizable images. In Proceedings of the IEEE Con- ference on Computer Vision and Pattern Recognition, pages 427–436, 2015. 1

  21. [29]

    Conjnorm: Tractable density estimation for out-of- distribution detection

    Bo Peng, Yadan Luo, Yonggang Zhang, Yixuan Li, and Zhen Fang. Conjnorm: Tractable density estimation for out-of- distribution detection. In The Twelfth International Confer- ence on Learning Representations, 2024. 2

  22. [30]

    Faster r-cnn: towards real-time object detection with region proposal networks

    Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster r-cnn: towards real-time object detection with region proposal networks. In Proceedings of the 28th International Conference on Neural Information Processing Systems - Vol- ume 1, page 91–99, Cambridge, MA, USA, 201...

  23. [31]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10684–10695, 2022. 2, 5

  24. [32]

    Berg, and Li Fei-Fei

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, San- jeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, Alexander C. Berg, and Li Fei-Fei. ImageNet Large Scale Visual Recognition Chal- lenge. International Journal of Computer Vision ...

  25. [33]

    A unified survey on anomaly, novelty, open-set, and out of-distribution detection: Solutions and future chal- lenges

    Mohammadreza Salehi, Hossein Mirzaei, Dan Hendrycks, Yixuan Li, Mohammad Hossein Rohban, and Mohammad Sabokrou. A unified survey on anomaly, novelty, open-set, and out of-distribution detection: Solutions and future chal- lenges. Transactions on Machine Learning Research, 2022. 1

  26. [34]

    First-order ad- versarial vulnerability of neural networks and input dimen- sion

    Carl-Johann Simon-Gabriel, Yann Ollivier, Leon Bottou, Bernhard Sch ¨olkopf, and David Lopez-Paz. First-order ad- versarial vulnerability of neural networks and input dimen- sion. In Proceedings of the 36th International Conference on Machine Learning, pages 5809–5817. PMLR, 2019. 4

  27. [35]

    Non- parametric outlier synthesis

    Leitian Tao, Xuefeng Du, Jerry Zhu, and Yixuan Li. Non- parametric outlier synthesis. In The Eleventh International Conference on Learning Representations, 2023. 2, 3

  28. [36]

    The inaturalist species classification and detection dataset

    Grant Van Horn, Oisin Mac Aodha, Yang Song, Yin Cui, Chen Sun, Alex Shepard, Hartwig Adam, Pietro Perona, and Serge Belongie. The inaturalist species classification and detection dataset. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages ...

  29. [37]

    Ehinger, Aude Oliva, and Antonio Torralba

    Jianxiong Xiao, James Hays, Krista A. Ehinger, Aude Oliva, and Antonio Torralba. Sun database: Large-scale scene recognition from abbey to zoo. In 2010 IEEE Computer So- ciety Conference on Computer Vision and Pattern Recogni- tion, pages 3485–3492, 2010. 5

  30. [38]

    Self-training with noisy student improves imagenet clas- sification

    Qizhe Xie, Minh-Thang Luong, Eduard Hovy, and Quoc V Le. Self-training with noisy student improves imagenet clas- sification. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 10687– 10698, 2020. 3

  31. [39]

    Generalized out-of-distribution detection: A survey

    Jingkang Yang, Kaiyang Zhou, Yixuan Li, and Ziwei Liu. Generalized out-of-distribution detection: A survey. Inter- national Journal of Computer Vision, pages 1–28, 2024. 1

  32. [40]

    Lsun: Construction of a large-scale image dataset using deep learning with humans in the loop

    Fisher Yu, Yinda Zhang, Shuran Song, Ari Seff, and Jianx- iong Xiao. Lsun: Construction of a large-scale image dataset using deep learning with humans in the loop. arXiv preprint arXiv:1506.03365, 2015. 5

  33. [41]

    Wide residual networks for mitosis detection

    Erwan Zerhouni, D ´avid L ´anyi, Matheus Viana, and Maria Gabrani. Wide residual networks for mitosis detection. In 2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017), pages 924–928. IEEE, 2017. 5

  34. [42]

    Places: An image database for deep scene understand- ing

    Bolei Zhou, Agata Lapedriza, Antonio Torralba, and Aude Oliva. Places: An image database for deep scene understand- ing. Journal of Vision, 17(10):296–296, 2017. 5 10 FEVER-OOD: Free Energy Vulnerability Elimination for Robust Out-of-Distribution Detection Supplementary Material

  35. [44]

    Mathematical Background In this section we provide a detailed analysis of the mathe- matical formulation of our methods, as described in Sec. 4. Specifically, we analyze the constraints in Eqs. (10) and (11) and the solution to the minimization problem in Eq. (12). 8.1. Null S...

  36. [45]

    2Code is available at https : / / github

    Training Regime We follow the original training regime for each baseline method and their corresponding FEVER-OOD variants2. 2Code is available at https : / / github . com / KostadinovShalon/fever-ood. 11 VOS [6]: we train all our VOS for classification models for 100 epochs w...

  37. [46]

    5 and 6 show the feature projections of VOS and FEVER-OOD VOS models using UMAP and t-SNE pro- jections, respectively

    Null Space Projection Figs. 5 and 6 show the feature projections of VOS and FEVER-OOD VOS models using UMAP and t-SNE pro- jections, respectively. Both models are for CIFAR-10 as in- distribution data, with the best model of FEVER-OOD VOS being shown, corresponding to an 96-NS...

  38. [47]

    9 to 13 show the ablation studies of varyingλLSV and λCN in Eqs

    Ablation Studies Figs. 9 to 13 show the ablation studies of varyingλLSV and λCN in Eqs. (13) and (14) for different classification meth- 12 (a) (b) (c) Figure 5. Feature Space UMAP Projection for models trained on CIFAR-10. Top row corresponds to the VOS [6] model while the bo...

  39. [48]

    14 for OpenImages [20] as OOD, and in Fig

    Qualitative Results Additional qualitative examples for object-level OOD de- tection using VOS [6] and FFS [19] models trained with and without FEVER-OOD with PASCAL VOC as in- distribution are shown in Fig. 14 for OpenImages [20] as OOD, and in Fig. 15 fos MS-COCO [25] as OOD

  40. [49]

    We identify the following limitations: • FEVER-OOD does not entirely avoid the null space vul- nerabilities

    Limitations and Potential Negative Impact Finally, this section discusses some limitations and potential negative impact of FEVER-OOD. We identify the following limitations: • FEVER-OOD does not entirely avoid the null space vul- nerabilities. While we reduce the size of it, t...

  41. [2022]

    1, 2, 3, 4, 5, 8, 12, 13, 15, 17, 18

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.