Pith. sign in

REVIEW 5 major objections 5 minor 58 references

Non-Linear Outlier Synthesis for Out-of-Distribution Detection

T0 review · 5 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read NCIS sets new out-of-distribution detection records by generating outliers in a diffusion model's conditioning space and fitting class-conditional manifolds with a volume-preserving network.

desk verdict Solid empirical outlier-synthesis paper with real gains and honest ablations; the main risks are borrowed baselines and test-set hyperparameter tuning, not the unvalidated geometry. read the letter →

arxiv 2411.13619 v1 pith:UYK5WHEY submitted 2024-11-20 cs.CV cs.AIcs.LG

Lars Doorenbos , Raphael Sznitman , Pablo Márquez-Neila This is my paper · ORCID
classification cs.CVcs.AIcs.LG
keywords out-of-distributiondetectiondiffusionmodelssyntheticoutliersynthesisconditionalvolume-preservingnetworkenergy-basedOODscoringStableCIFAR-100ImageNet-100
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

NCIS aims to make synthetic outlier images that sit right on the boundary of a classifier's known classes, so training on them teaches the classifier to flag unexpected inputs instead of confidently mislabeling them. The method embeds each training image directly in the diffusion model's conditioning space by a few gradient steps of the denoising objective, then fits class-conditional manifolds with a conditional volume-preserving network (cVPN). Sampling low-likelihood regions of those fitted distributions and decoding with Stable Diffusion produces near-OOD images. The paper reports this lowers the average false-positive rate at 95% true-positive rate (FPR95) by 29.47 on CIFAR-100 and by 4.87 on ImageNet-100 relative to the best previous methods. It also notes that the frozen Stable Diffusion decoder limits direct use in domains such as medical imaging.

What carries the argument

The load-bearing machinery is the combination of diffusion-space embeddings and a conditional volume-preserving network (cVPN). The embedding step treats each image's condition vector as a learnable token and takes three gradient steps of $\| \epsilon - \epsilon_\theta(\hat{x}_t,t,e)\|^2$ from the class label embedding, so the diffusion model itself defines the in-distribution region. The cVPN is a bijective, volume-preserving map $$f: \mathbb{R}^D \times \mathcal{Y} \to \mathbb{R}^D$$ composed of orthogonal and conditional coupling layers; its first $K$ outputs are invariants that stay near zero for in-class embeddings, and the remaining outputs carry within-class variability. Because the Jacobian has unit determinant, fitting a Gaussian to $v_i = f(e_i, y_i)$ and inverting gives a non-linear class-conditional density in diffusion space, and rejection sampling its low-likelihood tail yields embeddings that decode into boundary images.

What would settle it

Train the same classifier with outliers generated by the cVPN and with outliers from a plain Gaussian fitted directly in diffusion space, holding all other hyperparameters fixed; if the FPR95 gap is not attributable to the cVPN, the non-linear manifold claim is unsupported. A second decisive check is whether the three-step diffusion embeddings of held-out ID images sit closer to their own class token than to other class tokens; if they do not, the conditioning space is not semantically aligned as assumed.

Watch

Extended reading notes

Core claim

The central claim is that outlier synthesis for OOD detection should happen inside the diffusion model's own conditioning space, and that the in-distribution region there is class-conditional and non-linear. NCIS derives an embedding for each training image by minimizing the diffusion noise-prediction loss with respect to a condition vector initialized at the class token, so the collection of embeddings traces where the generator would produce in-distribution images. A conditional volume-preserving network then maps each class's embeddings to an invariant space where in-class points collapse toward zero, and class-conditional Gaussians fitted there induce non-linear densities back in diffusion space. Outliers are obtained by rejection-sampling low-likelihood vectors, inverting the network, and prompting Stable Diffusion; these images are used with an energy regularizer to train the classifier. Used this way, the synthesized near-OOD images are claimed to outperform previous synthetic-outlier pipelines on CIFAR-100 and ImageNet-100, with ablations attributing the improvement to each of the design choices.

Load-bearing premise

The load-bearing premise is that Stable Diffusion's conditioning space is semantically structured, so a few gradient steps put each image's embedding near its class label and low-likelihood points in that space decode to images just outside the class rather than to random corruptions.

Editorial extensions

If this is right

  • On CIFAR-100, NCIS lowers the average FPR95 to 10.84, a 29.47-point improvement over the best previous method, with classification accuracy largely unchanged.
  • On ImageNet-100, NCIS lowers the average FPR95 to 33.89, a 4.87-point improvement, and beats Dream-OOD on every OOD test set in that benchmark.
  • The ablation shows each component matters: diffusion-space embeddings improve on Dream-OOD's embeddings, the cVPN improves on Gaussian fitting, and matching the interpolation method between training and test time removes a large spurious drop.
  • The number of synthetic outliers follows a power law, so generating more outliers (up to 51,200) is a direct and large lever on detection performance.
  • The method transfers to different classifier backbones (ResNet, ViT, ConvNeXt), with bigger backbones giving better OOD detection as well as higher accuracy.

Reading between the lines

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

  • If the diffusion embedding is what carries the gain, the same recipe should transfer to other text-conditioned diffusion models: swap Stable Diffusion for a domain-specific generator and keep the cVPN module, which would address the medical-imaging limitation the authors state.
  • The interpolation-mismatch finding suggests a testable extension: perturbing low-level statistics of ID images during training (resizing, JPEG compression) could push the learned boundary toward semantics rather than texture, a direction the paper only begins to explore.
  • Because only three gradient steps are used per image, the embedding cost scales linearly with dataset size; the paper's subsampling experiment indicates a small fraction of ID embeddings suffices, so the method should scale to much larger training sets.
  • A stricter test of the boundary claim would be to measure how often cVPN-sampled outliers are semantically related to their conditioning class; if they are mostly unrelated, the gain might come from the energy regularization rather than from precise boundary alignment.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper proposes NCIS, a method for out-of-distribution (OOD) detection that generates synthetic outliers using Stable Diffusion. Each training image is embedded into the diffusion conditioning space by taking three gradient steps of the noise-prediction loss starting from the class label embedding (Algorithm 1, Eq. (1)). A class-conditional volume-preserving network (cVPN) maps these embeddings into an invariant space where class-conditional Gaussians are fitted (Eqs. (4), (8)). Low-likelihood samples are drawn from these Gaussians, mapped back through the inverse cVPN, and rendered by Stable Diffusion into synthetic OOD images, which are then used to regularize a classifier with an energy-based OOD loss (Eqs. (12)-(13)). Experiments on CIFAR-100 and ImageNet-100 report state-of-the-art FPR95/AUC, with ablations of the embedding, the sampling distribution, the resizing strategy, and hyperparameters, plus experiments across architectures and outlier counts.

Significance. If the empirical claims hold under independent validation, this is a strong contribution to synthetic-outlier OOD detection: the reported gains over Dream-OOD are large on CIFAR-100, the method generalizes across architectures, the code is publicly available, and the analysis of resizing/interpolation sensitivity is a useful and somewhat surprising finding. The cVPN contribution is a natural extension of the authors' prior NL-Invs work and is clearly motivated. However, the central state-of-the-art claim currently rests on test-set hyperparameter selection and on copied baseline numbers without variance, and the geometric assumption that low-likelihood invariant-space samples map to useful near-OOD images is not quantitatively validated.

major comments (5)
  1. [Section 5.2, Fig. 6] The hyperparameters lambda and p are selected by evaluating FPR95 on the same CIFAR-100 OOD test sets that are used in the main comparison of Table 1. Since lambda directly controls the 'degree of out-of-distributionness' of the generated outliers (Section 3.3), tuning it on the test OOD distribution can substantially inflate the reported average FPR95 of 10.84. Please either select these hyperparameters on a held-out validation split and then evaluate once on the test benchmarks, or report results for a fixed default choice across all benchmarks, clearly stating which numbers are the result of test-set selection.
  2. [Tables 1 and 2] All baseline numbers are copied from Dream-OOD [12] with no error bars, while NCIS results are averaged over three seeds. On ImageNet-100 the reported improvement over Dream-OOD is only 4.87 FPR95 (33.89 vs. 38.76), which is within the plausible run-to-run variance of a single training run. The claim of state-of-the-art on ImageNet-100 therefore needs either multiple seeds for the baselines under the same codebase, or at least a statement of Dream-OOD's reported variance if it exists in [12].
  3. [Section 3.4, Fig. 5] The method's core assumption is that a low-likelihood point v' in the invariant space, when mapped back through the inverse cVPN, lands near the ID/OOD boundary of the diffusion conditioning space. However, Eq. (4) only constrains the forward map on ID embeddings; the inverse map f^{-1} is unconstrained outside the training support. The paper provides only qualitative examples (Fig. 5) and indirect ablations (Table 3) as evidence. Please add a quantitative validation of the near-boundary property, for example by measuring the distance of generated embeddings e' to the ID embedding manifold, or by comparing the effect of NCIS-generated outliers with that of random corruptions or random perturbations of ID embeddings under the same resizing and training protocol.
  4. [Section 3.1, Algorithm 1] The text calls the embedding e a maximum-a-posteriori estimate, but the algorithm initializes e at the label embedding and performs only three gradient steps on the first term of Eq. (1). No evidence is provided that three steps approximate the MAP solution, nor is the iteration count ablated. Since the quality of all downstream modeling depends on these embeddings, please either show convergence behavior (e.g., loss curves or embedding drift over steps) or explicitly state that three steps is a heuristic and measure its sensitivity.
  5. [Section 3.4] The rejection sampling procedure is not specified: the paper says 'we apply rejection sampling in the invariant space to obtain an outlier v' from the low-likelihood regions' but does not define the acceptance threshold, the quantile, or the number of samples tried. This is a load-bearing step of the generation pipeline and must be described precisely for reproducibility.
minor comments (5)
  1. [Eq. (1)] The regularization term R(e, ey) appears in the objective but is never explicitly defined; the text later states it is implemented by initialization to ey, but the form of R should be stated or the notation should be removed to avoid confusion.
  2. [Table 3] The row label 'Both cVPN Ours' is cryptic; please clarify that 'Both' refers to combining both resizing augmentations with the cVPN-based sampling.
  3. [Section 5.2] The sentence 'all other baselines are unaffected by the interpolation method' is asserted without supporting evidence; if retained, please provide the corresponding experiment or cite a table.
  4. [Table 4] The per-architecture results are reported without error bars or the number of seeds; adding these would strengthen the claim of robustness across architectures.
  5. [Fig. 6(b)] The x-axis label 'p (number of invariants)' is confusing because p is described earlier as a variance threshold; please clarify the relationship between p and the resulting number of invariants K.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the NCIS derivation is self-contained and the benchmark claim is evaluated on external OOD data.

full rationale

The paper's derivation chain is: fit diffusion embeddings e by optimizing Eq. (1) for each ID image; train the conditional volume-preserving network with Eq. (4) to map each class's embeddings to near-zero invariants; fit a class-conditional Gaussian (Eq. (8)) in invariant space; sample low-likelihood v' and invert through f^{-1} (Sec. 3.4) to obtain diffusion-space prompts; render synthetic outliers with Stable Diffusion; and finally regularize the classifier with Eq. (13). No equation in this chain takes the benchmark FPR95/AUC numbers as an input, and no fitted parameter is renamed as a prediction: the synthetic outliers are generated from the fitted ID density, while the state-of-the-art claim is tested on external OOD datasets (SVHN, Places365, LSUN, iSun, Textures, iNaturalist, SUN) that are not used in the embedding, cVPN, or Gaussian fitting. The self-citations [9,10] supply the non-linear invariant/VPN architecture and are acknowledged as prior work; the paper defines its conditional coupling layers in Eqs. (6)-(7) and reports ablations (Table 3) that isolate the cVPN's contribution, so the argument does not reduce to an unverified assertion in those citations. The near-OOD boundary behavior of f^{-1} (the weakest assumption) is an empirical validity question about whether Stable Diffusion renders useful boundary images; it is not circularity, because the final evaluation is independent of that assumption's truth. The stated limitation about the frozen decoder is a scope restriction for medical imaging, not a circular step.

Assumptions & free parameters 5 free parameters · 4 assumptions · 1 invented entities

The method relies on several hand-chosen hyperparameters (iterations, lambda, p, beta) and on domain assumptions about the semantic alignment of Stable Diffusion's conditioning space. The cVPN is a new architecture built on the authors' prior NL-Invs work.

free parameters (5)
  • Embedding optimization iterations = 3
    Chosen by hand; no convergence analysis provided.
  • Regularization strength lambda = 1e-5
    Tuned on CIFAR-100 benchmark (Fig. 6a).
  • Variance threshold p = 2
    Tuned on CIFAR-100 benchmark (Fig. 6b).
  • Beta for OOD loss = 1.0
    Fixed without ablation.
  • Embedding learning rate eta = Not reported
    Hyperparameter for gradient descent in Algorithm 1 is not disclosed, hindering reproducibility.
assumptions (4)
  • domain assumption Stable Diffusion v1.4's conditioning space is semantically aligned so that points near a class label embedding produce images of that class.
    Used in Section 3.1 and 3.4 to justify that embeddings and sampled points correspond to meaningful images.
  • domain assumption The cVPN from [10] remains bijective when made conditional, and its Jacobian determinant is exactly 1.
    Stated in Section 3.2-3.3, relying on the volume-preserving construction from [10] without formal proof in this paper.
  • ad hoc to paper Three gradient steps of Eq. (1) suffice to approximate the MAP embedding.
    Algorithm 1 sets iterations to 3 without justification; the MAP interpretation is therefore approximate.
  • ad hoc to paper A Gaussian in invariant space, after rejection sampling at low likelihood, yields useful OOD samples.
    Section 3.3-3.4 assumes the cVPN-transformed distribution is well-modeled by a Gaussian and that low-likelihood points map to near-boundary outliers.
invented entities (1)
  • Conditional volume-preserving network (cVPN) independent evidence
    purpose: Fit class-conditional manifolds in the diffusion embedding space and enable non-linear density estimation.
    New network component introduced in this paper, evaluated empirically on external OOD benchmarks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Non-Linear Outlier Synthesis for Out-of-Distribution Detection." pith.science (2026). https://pith.science/paper/UYK5WHEY

@misc{pith2026241113619,
  author       = {Pith},
  title        = {Pith review of: Non-Linear Outlier Synthesis for Out-of-Distribution Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UYK5WHEY}},
  note         = {Machine review of arXiv:2411.13619}
}
read the original abstract

The reliability of supervised classifiers is severely hampered by their limitations in dealing with unexpected inputs, leading to great interest in out-of-distribution (OOD) detection. Recently, OOD detectors trained on synthetic outliers, especially those generated by large diffusion models, have shown promising results in defining robust OOD decision boundaries. Building on this progress, we present NCIS, which enhances the quality of synthetic outliers by operating directly in the diffusion's model embedding space rather than combining disjoint models as in previous work and by modeling class-conditional manifolds with a conditional volume-preserving network for more expressive characterization of the training distribution. We demonstrate that these improvements yield new state-of-the-art OOD detection results on standard ImageNet100 and CIFAR100 benchmarks and provide insights into the importance of data pre-processing and other key design choices. We make our code available at \url{https://github.com/LarsDoorenbos/NCIS}.

Figures

Figures reproduced from arXiv: 2411.13619 by the authors.

Figure 1
Figure 1. Random outliers generated for three CIFAR100 classes by our method. Using our generated samples as auxil￾iary outliers during training greatly improves the OOD detection performance of modern classifiers. A major difficulty in doing so is the lack of real OOD samples at training time. Recent advancements in super￾vised OOD detection tackle this challenge by generating synthetic OOD samples and using them during trai… view at source ↗
Figure 2
Figure 2. Comparison between (a) Dream-OOD, (b) linear invariants on our embeddings, and (c) our proposed method on a toy example. The disjoint embeddings and normalization of Dream-OOD greatly limit the flexibility of the generated outliers. Linear invariants similarly lack capacity. On the other hand, our method can generate successful outliers by modeling arbitrary distributions. Finally, we make heavy use of diffusion mod… view at source ↗
Figure 3
Figure 3. The class-specific representations learned by the cVPN on toy data with three classes. Depending on the conditioning, the cVPN transforms the input data such that the current class is transformed into a representation with an invariant (the y-axis). The background color indicates the distance to the nearest training data point from the current class in the original space. Images with a white￾shaded background result… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Effect of regularization on generated outliers. Out￾liers get progressively more OOD with stronger regularization, providing an intuitive way to control their difficulty. 3.4. OOD sample generation Generating an OOD image for a given class y follows nat￾urally from the…
Figure 5
Figure 5. Figure 5: Nine random generated outliers by our method and Dream-OOD for CIFAR100 (top) and ImageNet-100 (bottom). Our generated outliers are closer to the intended meaning, provid￾ing better signal to learn the ID/OOD decision boundary. 5.1. Qualitative examples [PITH_FULL_IMA…
Figure 6
Figure 6. Figure 6: Effect of λ (a) and p (b) on the FPR95 (↓) with CI￾FAR100 as the in-distribution. Difficult samples (see [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

58 extracted references · 42 canonical work pages

  1. [12]

    Dream the impossible: Outlier imagination with diffusion models

    Xuefeng Du, Yiyou Sun, Jerry Zhu, and Yixuan Li. Dream the impossible: Outlier imagination with diffusion models. Advances in Neural Information Processing Systems , 36,

  2. [1]

    Latent space autoregression for novelty detec- tion

    Davide Abati, Angelo Porrello, Simone Calderara, and Rita Cucchiara. Latent space autoregression for novelty detec- tion. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition, pages 481–490, 2019. 2

    work page 2019

  3. [2]

    Detecting semantic anomalies

    Faruk Ahmed and Aaron Courville. Detecting semantic anomalies. In Proceedings of the AAAI Conference on Ar- tificial Intelligence, pages 3154–3162, 2020. 2

    work page 2020

  4. [3]

    Gradorth: A simple yet efficient out- of-distribution detection with orthogonal projection of gra- dients

    Sima Behpour, Thang Long Doan, Xin Li, Wenbin He, Liang Gou, and Liu Ren. Gradorth: A simple yet efficient out- of-distribution detection with orthogonal projection of gra- dients. Advances in Neural Information Processing Systems, 36, 2024. 2

    work page 2024

  5. [4]

    Fodfom: Fake outlier data by founda- tion models creates stronger visual out-of-distribution detec- tor

    Jiankang Chen, Ling Deng, Zhiyong Gan, Wei-Shi Zheng, and Ruixuan Wang. Fodfom: Fake outlier data by founda- tion models creates stronger visual out-of-distribution detec- tor. In ACM Multimedia 2024, 2024. 1, 2

    work page 2024

  6. [5]

    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,

  7. [6]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 248–255,

    work page

  8. [7]

    Learning confi- dence for out-of-distribution detection in neural networks

    Terrance DeVries and Graham W Taylor. Learning confi- dence for out-of-distribution detection in neural networks. arXiv preprint arXiv:1802.04865, 2018. 2

    arXiv 2018

Show all 58 references
  1. [8]

    Diffusion models beat gans on image synthesis

    Prafulla Dhariwal and Alexander Nichol. Diffusion models beat gans on image synthesis. Advances in neural informa- tion processing systems, 34:8780–8794, 2021. 3

    2021

  2. [9]

    Data invariants to understand unsupervised out-of- distribution detection

    Lars Doorenbos, Raphael Sznitman, and Pablo M ´arquez- Neila. Data invariants to understand unsupervised out-of- distribution detection. In European Conference on Computer Vision, pages 133–150. Springer, 2022. 2, 4

    2022

  3. [10]

    Learning non-linear invariants for unsupervised out- of-distribution detection

    Lars Doorenbos, Raphael Sznitman, and Pablo M ´arquez- Neila. Learning non-linear invariants for unsupervised out- of-distribution detection. 2024. 2, 4, 5, 6

    2024

  4. [11]

    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 Proceedings of the International Conference on Learning Representations, 2022. 1, 2, 5, 6, 7

    2022

  5. [13]

    Ex- ploring the limits of out-of-distribution detection

    Stanislav Fort, Jie Ren, and Balaji Lakshminarayanan. Ex- ploring the limits of out-of-distribution detection. Advances in Neural Information Processing Systems , 34:7068–7081,

  6. [14]

    Transfusion–a transparency-based diffusion model for anomaly detection

    Matic Fu ˇcka, Vitjan Zavrtanik, and Danijel Sko ˇcaj. Transfusion–a transparency-based diffusion model for anomaly detection. In European conference on computer vi- sion, pages 91–108. Springer, 2025. 3

    2025

  7. [15]

    An image is worth one word: Personalizing text-to-image gen- eration using textual inversion

    Rinon Gal, Yuval Alaluf, Yuval Atzmon, Or Patashnik, Amit H Bermano, Gal Chechik, and Daniel Cohen-Or. An image is worth one word: Personalizing text-to-image gen- eration using textual inversion. Proceedings of the Interna- tional Conference on Learning Representations, 2023. 3

    2023

  8. [16]

    Hierarchical vaes know what they don’t know

    Jakob D Havtorn, Jes Frellsen, Søren Hauberg, and Lars Maaløe. Hierarchical vaes know what they don’t know. InIn- ternational Conference on Machine Learning , pages 4117–

  9. [17]

    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. Proceedings of the International Conference on Learning Representations, 2017. 1, 2, 6, 7

    2017

  10. [18]

    Deep anomaly detection with outlier exposure

    Dan Hendrycks, Mantas Mazeika, and Thomas Dietterich. Deep anomaly detection with outlier exposure. International Conference on Learning Representations, 2019. 2

    2019

  11. [19]

    Using self-supervised learning can improve model robustness and uncertainty

    Dan Hendrycks, Mantas Mazeika, Saurav Kadavath, and Dawn Song. Using self-supervised learning can improve model robustness and uncertainty. Advances in Neural In- formation Processing Systems, 32, 2019. 2

    2019

  12. [20]

    Scal- ing out-of-distribution detection for real-world settings

    Dan Hendrycks, Steven Basart, Mantas Mazeika, Moham- madreza Mostajabi, Jacob Steinhardt, and Dawn Song. Scal- ing out-of-distribution detection for real-world settings. In- ternational Conference on Machine Learning, 2022. 2

    2022

  13. [21]

    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. 2, 6, 7

    2020

  14. [22]

    Mos: Towards scaling out-of- distribution detection for large semantic space

    Rui Huang and Yixuan Li. Mos: Towards scaling out-of- distribution detection for large semantic space. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8710–8719, 2021. 1

    2021

  15. [23]

    On the impor- tance of gradients for detecting distributional shifts in the wild

    Rui Huang, Andrew Geng, and Yixuan Li. On the impor- tance of gradients for detecting distributional shifts in the wild. Advances in Neural Information Processing Systems , 34:677–689, 2021. 2

    2021

  16. [24]

    Why is the mahalanobis distance effective for anomaly detection? arXiv preprint arXiv:2003.00402, 2020

    Ryo Kamoi and Kei Kobayashi. Why is the mahalanobis distance effective for anomaly detection? arXiv preprint arXiv:2003.00402, 2020. 2

    2003 arXiv

  17. [25]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny images. 2009. 6, 1

    2009

  18. [26]

    Training confidence-calibrated classifiers for detecting out- of-distribution samples

    Kimin Lee, Honglak Lee, Kibok Lee, and Jinwoo Shin. Training confidence-calibrated classifiers for detecting out- of-distribution samples. International Conference on Learn- ing Representations, 2018. 2

    2018

  19. [27]

    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. 2, 6, 7

    2018

  20. [28]

    Cutpaste: Self-supervised learning for anomaly de- tection and localization

    Chun-Liang Li, Kihyuk Sohn, Jinsung Yoon, and Tomas Pfister. Cutpaste: Self-supervised learning for anomaly de- tection and localization. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 9664–9674, 2021. 2

    2021

  21. [29]

    Enhanc- ing the reliability of out-of-distribution image detection in 9 neural networks

    Shiyu Liang, Yixuan Li, and Rayadurgam Srikant. Enhanc- ing the reliability of out-of-distribution image detection in 9 neural networks. In Proceedings of the International Con- ference on Learning Representations, 2018. 2, 6, 7

    2018

  22. [30]

    Energy-based out-of-distribution detection

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

  23. [31]

    Gen: Pushing the limits of softmax-based out-of-distribution de- tection

    Xixi Liu, Yaroslava Lochman, and Christopher Zach. Gen: Pushing the limits of softmax-based out-of-distribution de- tection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 23946– 23955, 2023. 2

    2023

  24. [32]

    Do deep gener- ative models know what they don’t know? arXiv preprint arXiv:1810.09136, 2018

    Eric Nalisnick, Akihiro Matsukawa, Yee Whye Teh, Di- lan Gorur, and Balaji Lakshminarayanan. Do deep gener- ative models know what they don’t know? arXiv preprint arXiv:1810.09136, 2018. 2

    2018 arXiv

  25. [33]

    Reading digits in natural images with unsupervised feature learning

    Yuval Netzer, Tao Wang, Adam Coates, Alessandro Bis- sacco, Bo Wu, and Andrew Y Ng. Reading digits in natural images with unsupervised feature learning. 2011. 6, 1

    2011

  26. [34]

    Outlier exposure with confidence control for out-of-distribution detection

    Aristotelis-Angelos Papadopoulos, Mohammad Reza Rajati, Nazim Shaikh, and Jiamian Wang. Outlier exposure with confidence control for out-of-distribution detection. Neuro- computing, 441:138–150, 2021. 2

    2021

  27. [35]

    Mean-shifted contrastive loss for anomaly detection

    Tal Reiss and Yedid Hoshen. Mean-shifted contrastive loss for anomaly detection. In Proceedings of the AAAI Confer- ence on Artificial Intelligence, pages 2155–2162, 2023. 2

    2023

  28. [36]

    A simple fix to mahalanobis distance for improving near-ood detection

    Jie Ren, Stanislav Fort, Jeremiah Liu, Abhijit Guha Roy, Shreyas Padhy, and Balaji Lakshminarayanan. A simple fix to mahalanobis distance for improving near-ood detection. arXiv preprint arXiv:2106.09022, 2021. 2

    2021 arXiv

  29. [37]

    High-resolution image syn- thesis with latent diffusion models, 2021

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj¨orn Ommer. High-resolution image syn- thesis with latent diffusion models, 2021. 3

    2021

  30. [38]

    Detecting out-of-distribution examples with gram matrices

    Chandramouli Shama Sastry and Sageev Oore. Detecting out-of-distribution examples with gram matrices. In Interna- tional Conference on Machine Learning , pages 8491–8501. PMLR, 2020. 2

    2020

  31. [39]

    Understanding anomaly detection with deep invert- ible networks through hierarchies of distributions and fea- tures

    Robin Schirrmeister, Yuxuan Zhou, Tonio Ball, and Dan Zhang. Understanding anomaly detection with deep invert- ible networks through hierarchies of distributions and fea- tures. Advances in Neural Information Processing Systems , 33:21038–21049, 2020. 2, 7

    2020

  32. [40]

    Natural synthetic anomalies for self-supervised anomaly detection and localization

    Hannah M Schl ¨uter, Jeremy Tan, Benjamin Hou, and Bern- hard Kainz. Natural synthetic anomalies for self-supervised anomaly detection and localization. In European Conference on Computer Vision, pages 474–489. Springer, 2022. 2

    2022

  33. [41]

    Ssd: A unified framework for self-supervised outlier detection

    Vikash Sehwag, Mung Chiang, and Prateek Mittal. Ssd: A unified framework for self-supervised outlier detection. In- ternational Conference on Learning Representations , 2021. 2, 3

    2021

  34. [42]

    Input complexity and out-of- distribution detection with likelihood-based generative mod- els

    Joan Serr `a, David ´Alvarez, Vicenc ¸ G´omez, Olga Slizovskaia, Jos´e F N´u˜nez, and Jordi Luque. Input complexity and out-of- distribution detection with likelihood-based generative mod- els. International Conference on Learning Representations,

  35. [43]

    Dice: Leveraging sparsification for out-of-distribution detection

    Yiyou Sun and Yixuan Li. Dice: Leveraging sparsification for out-of-distribution detection. InProceedings of European Conference on Computer Vision, 2022. 6, 7

    2022

  36. [44]

    React: Out-of- distribution detection with rectified activations

    Yiyou Sun, Chuan Guo, and Yixuan Li. React: Out-of- distribution detection with rectified activations. In Advances in Neural Information Processing Systems, 2021. 6, 7

    2021

  37. [45]

    Out-of- distribution detection with deep nearest neighbors

    Yiyou Sun, Yifei Ming, Xiaojin Zhu, and Yixuan Li. Out-of- distribution detection with deep nearest neighbors. In Pro- ceedings of the International Conference on Machine Learn- ing, pages 20827–20840, 2022. 2, 6, 7, 1

    2022

  38. [46]

    Detecting outliers with poisson image interpolation

    Jeremy Tan, Benjamin Hou, Thomas Day, John Simpson, Daniel Rueckert, and Bernhard Kainz. Detecting outliers with poisson image interpolation. In Medical Image Com- puting and Computer Assisted Intervention–MICCAI 2021: 24th International Conference, Strasbourg, France, Septem-...

    2021

  39. [47]

    Non- parametric outlier synthesis

    Leitian Tao, Xuefeng Du, Xiaojin Zhu, and Yixuan Li. Non- parametric outlier synthesis. In Proceedings of the Interna- tional Conference on Learning Representations, 2023. 1, 2, 5, 6, 7

    2023

  40. [48]

    Self-guided generation of minority samples using diffusion models

    Soobin Um and Jong Chul Ye. Self-guided generation of minority samples using diffusion models. 2024. 3

    2024

  41. [49]

    The inaturalist species classification and de- tection 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 de- tection dataset. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , ...

    2018

  42. [50]

    Vim: Out-of-distribution with virtual-logit matching

    Haoqi Wang, Zhizhong Li, Litong Feng, and Wayne Zhang. Vim: Out-of-distribution with virtual-logit matching. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4921–4930, 2022. 6, 7

    2022

  43. [51]

    Contrastive training for improved out- of-distribution detection

    Jim Winkens, Rudy Bunel, Abhijit Guha Roy, Robert Stanforth, Vivek Natarajan, Joseph R Ledsam, Patricia MacWilliams, Pushmeet Kohli, Alan Karthikesalingam, Si- mon Kohl, et al. Contrastive training for improved out- of-distribution detection. arXiv preprint arXiv:2007.05566,

    2007 arXiv

  44. [52]

    Datasetdm: Synthesizing data with perception annota- tions using diffusion models

    Weijia Wu, Yuzhong Zhao, Hao Chen, Yuchao Gu, Rui Zhao, Yefei He, Hong Zhou, Mike Zheng Shou, and Chunhua Shen. Datasetdm: Synthesizing data with perception annota- tions using diffusion models. Advances in Neural Informa- tion Processing Systems, 36:54683–54695, 2023. 3

    2023

  45. [53]

    Sun database: Large-scale scene recognition from abbey to zoo

    Jianxiong Xiao, James Hays, Krista A Ehinger, Aude Oliva, and Antonio Torralba. Sun database: Large-scale scene recognition from abbey to zoo. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3485–3492, 2010. 6, 1

    2010

  46. [54]

    Do we really need to learn representations from in-domain data for outlier de- tection? ICML 2021 Workshop on Uncertainty & Robustness in Deep Learning, 2021

    Zhisheng Xiao, Qing Yan, and Yali Amit. Do we really need to learn representations from in-domain data for outlier de- tection? ICML 2021 Workshop on Uncertainty & Robustness in Deep Learning, 2021. 2

    2021

  47. [55]

    Turkergaze: Crowdsourcing saliency with webcam based eye tracking

    Pingmei Xu, Krista A Ehinger, Yinda Zhang, Adam Finkel- stein, Sanjeev R Kulkarni, and Jianxiong Xiao. Turkergaze: Crowdsourcing saliency with webcam based eye tracking. arXiv preprint arXiv:1504.06755, 2015. 6, 1

    2015 arXiv

  48. [56]

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

    Fisher Yu, Ari Seff, Yinda Zhang, Shuran Song, Thomas Funkhouser, and Jianxiong Xiao. Lsun: Construction of a large-scale image dataset using deep learning with humans in the loop. arXiv preprint arXiv:1506.03365, 2015. 6, 1 10

    2015 arXiv

  49. [57]

    Places: A 10 million image database for scene recognition

    Bolei Zhou, Agata Lapedriza, Aditya Khosla, Aude Oliva, and Antonio Torralba. Places: A 10 million image database for scene recognition. IEEE transactions on pattern analysis and machine intelligence, 40(6):1452–1464, 2017. 6, 1 11 Non-Linear Outlier Synthesis for Out-of-Distr...

    2017

  50. [58]

    The ID datasets are CIFAR-100 and ImageNet-100, which we briefly describe below: CIFAR-100 [25] contains 50’000 training images and 10’000 testing images belonging to 100 classes

    Dataset Details We follow the exact experimental protocol of [12]. The ID datasets are CIFAR-100 and ImageNet-100, which we briefly describe below: CIFAR-100 [25] contains 50’000 training images and 10’000 testing images belonging to 100 classes. ImageNet-100 is a subset of th...

Pith tools

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