betaFeedback

WorksComputer Science

Should We Pay This Much for Robustness? Efficient Proxy Certificates with Marginal Guarantees

Abstract

Modern classifiers remain sensitive to small input perturbations, while strong model-agnostic robustness certificates such as randomized smoothing are often too expensive for deployment, requiring an extensive number (e.g. 2000) of model forward passes per input. We propose a framework that delegates certification to a cheap proxy function while statistically controlling its deviation from the original sound certificate. Given an exchangeable unlabeled calibration set, we tune the proxy via robust conformal risk control to guarantee a bounded marginal (rather than pointwise) false-approval rate: the probability, over future inputs, that the proxy accepts an input whose prediction is flipped due to perturbation. The resulting proxy certificate can be used either as a standalone marginal certificate or as a fallback mechanism that invokes the expensive certificate only when the proxy rejects. We provide both single-sample and sample-heavy calibration recipes. Empirically, as a standalone certificate our method reduces the test-time certification cost to a single model forward while retaining much of the certified performance of randomized smoothing at small radii, and as a fallback it recovers the certified ratio of randomized smoothing at a more modest speedup. This enables cheap yet effective robustness guarantees for larger architectures such as vision transformers and language models.

Introduction

High accuracy alone is not sufficient for reliability in machine learning. Although modern neural networks achieve strong empirical performance, their predictions can be highly sensitive to small, often imperceptible, input perturbations [18]. This sensitivity can be exploited by adversaries to easily degrade the model’s performance. There are many empirical, and heuristic defenses proposed in the literature which have been circumvented by stronger or adaptive attacks [5]. A principled way to address this issue is to use robustness certificates, which provide formal guarantees that the model prediction remains invariant within a predefined perturbation set. Certificates can be viewed as binary decision rules: when they certify an input at a sufficiently large radius, they signal that the prediction can be trusted with respect to the specified threat model, whether the input is clean or even if adversarially perturbed, the perturbation is not effective.

Existing robustness certificates typically obtain such guarantees through architecture-specific verification methods, such as convex relaxations, or mixed-integer programming [38, 24, 34]. While powerful, these methods require white-box access to the model, and often need to be adapted when new architectures or layers are introduced. Alternatively, randomized smoothing (RS) [11] wraps around any black-box classifier, and convolves it with a noise augmentation. The resulting smoothed classifier changes slowly with local perturbations. Hence, one can derive closed-form model-agnostic bounds on its change. This black-box nature is a major practical advantage, allowing it to easily apply to large modern networks, and even to API-accessible systems [10]. However, randomized smoothing-based certificates require a large number of model forward passes (e.g. 2×1032 \times10^{3}) to estimate the smoothed class probabilities with high confidence, which can be prohibitively expensive especially with limited resources, or when the model itself is expensive to evaluate.

This cost is especially wasteful because not every input received by a system is adversarial; indeed, most inputs are expected to be clean. Moreover, in many settings, it is cheap to obtain a relatively small set of unlabeled samples representative of the deployment distribution. These observations motivate a distributional relaxation of robustness certification: rather than requiring a local certificate for every individual input, it may suffice to guarantee that certification failure is controllable below an adjustable low probability over test inputs.

In light of the arguments above, we define Proxy Certificate (ProxC), a framework for delegating expensive certification to cheap proxy functions under calibrated statistical control. Through a variant of conformal risk control (CRC) that we define in § 2.2, we control how often the proxy deviates from the expensive function. Our proxy reduces the cost of the certification step to a single model forward. We define three variants for ProxC, tailored for various compute budgets: ProxC-A, that uses a single forward pass per calibration point, ProxC which computes an expensive backbone certificate on the small calibration set, and ProxC-H which can increase the certified ratio, in return for more Monte-Carlo (MC) sampling. All three variants use a single sample to certify each test point. We bound the risk anchored at the latent clean point, which lets a sound certificate computed only on clean calibration points calibrate a cheap proxy that remains valid under perturbation (see § F).

The resulting proxy can be used either as a standalone certificate or as a hybrid fallback mechanism that only calls the expensive certification on inputs rejected by the proxy. These are two different operating regimes: the standalone proxy gives the largest saving, while the fallback recovers the certified ratio of the backbone at a smaller speedup. As we show in Table 1, by relaxing the soundness to bounded false approvals, we can reduce the cost of the certification step to a single forward, the same as verification-based approaches (importantly, the choice of label passed into the certificate is excluded from the cost evaluation. This can still follow the inference rule of the backbone). ProxC shows a higher certified accuracy compared to verification (note that our guarantee is marginal, and we use a slightly different architecture), and it is cheaper than randomized smoothing. In other words, in the trade-off between computational efficiency and applicability, we now propose a new dimension which is soundness. We propose smoothing-based guarantees with a realistic computational cost that is suitable for larger models like ViT, or language modeling in realistic time-sensitive scenarios.

MethodVariantGuaranteeCert. RatioCert. AccuracyError RateForwards / image
Rand. Smoothing—Pointwise70.06%56.08%——19502000
VerificationHu et al. [ [20]Pointwise—35.0%——11
ProxCStandaloneMarginal56.11%48.53%2.29%20001 (1950×)51 (39.2×)
ProxCFallbackMarginal72.35%57.3%2.29%2000779 (2.50×)829 (2.41×)

Table 1. Results on ImageNet/ResNet, r=0.14r=0.14, σ=0.5\sigma=0.5, α=0.05\alpha=0.05, Gaussian smoothing (2000 MC samples: 50 for the majority-vote label and 1950 for the certificate). The verification entry does not use smoothing (result directly reported from the reference); it gives a pointwise guarantee on a different architecture, so it is not directly comparable to the marginal guarantee of ProxC. Speedup is on forward calls and relative to randomized smoothing. Certification-only cost excludes the forwards for the label; end-to-end cost includes them.

Background

Problem setup. Let f:X→R∣Y∣f:\mathcal{X} \to\mathbb{R}^{|\mathcal{Y}|} be a pre-trained classifier (can be a single model, a pipeline of models, or any general inference policy, e.g. majority vote over random augmentations) that returns a score per class. For an input xi∈Xx_i \in\mathcal{X}, its top prediction is yi=arg⁡max⁡yif(xi)yiy_i = \arg\max_{y_i} f(x_i)_{y_i}. For brevity, whenever ff is compared across two inputs, e.g. f(x~)≠f(x)f(\tilde{x}) \ne f(x), we compare the predicted labels, i.e. arg⁡max⁡yf(x~)y≠arg⁡max⁡yf(x)y\arg\max_y f(\tilde{x})_y \ne\arg\max_y f(x)_y. For the test point, xn+1x_{n+1}, an adversary introduces a perturbation x~n+1=xn+1+δ\tilde{x}_{n+1}=x_{n+1}+\delta aiming to change the model’s prediction f(x~n+1)≠f(xn+1)f(\tilde{x}_{n+1}) \ne f(x_{n+1}). The adversary is limited to keep the perturbation imperceptible – x~n+1\tilde{x}_{n+1} has to be very close to xn+1x_{n+1}. This limitation is formally presented as a perturbation ball B:X→2X\mathcal{B} : \mathcal{X} \to2^{\mathcal{X}} – the set of all points from a given xx that the adversary is allowed to choose. A common example for images is the ℓ2\ell_2 ball Br(x)={x~:∥x~−x∥2≤r}\mathcal{B}_r(x) = \{\tilde{x} : \lVert\tilde{x} - x \rVert_2 \le r\}. The adversary has white-box access to the model (its architecture, parameters, etc), and the clean input xn+1x_{n+1}, but is not in control of the inference, and is unaware of the random seed used by the MC samples evaluated at test time. The test-time noise is freshly sampled for every query and is unknown when the adversary chooses its perturbation; hence we do not require a permanently secret global seed. We study the worst-case robustness: for any x~n+1∈Br(xn+1)\tilde{x}_{n+1} \in\mathcal{B}_r(x_{n+1}), the prediction should either remain the same, or the defense should abstain from certifying it. This is what we call the threat model.

Robustness certificate. A robustness certificate is a binary decision function

cert⁡f,r(xi)=1⇒∀x~∈Br(xi):f(x~)=f(xi).(1)\operatorname{cert}_{f,r}(x_i) = 1 \qquad\Rightarrow\qquad\forall\tilde{x} \in\mathcal{B}_r(x_i) : f(\tilde{x}) = f(x_i). \tag*{(1)}

The certificate is sound: cert⁡f,r(xi)=1\operatorname{cert}_{f,r}(x_i) = 1 ensures ff is stable entirely within Br(xi)\mathcal{B}_r(x_i), but it is not complete: cert⁡f,r(xi)=0\operatorname{cert}_{f,r}(x_i) = 0 does not necessarily imply vulnerability. The standard use case of the certificate is to apply it at test time on each (potentially perturbed) input x~n+1\tilde{x}_{n+1}, and if cert⁡f,r(x~n+1)=1\operatorname{cert}_{f,r}(\tilde{x}_{n+1}) = 1, we can safely use the prediction f(x~n+1)f(\tilde{x}_{n+1}); otherwise, we abstain.

Inverse ball B−1\mathcal{B}^{-1}. While we allow the adversary to perturb xx within B(x)\mathcal{B}(x), from the defender’s perspective, we should ensure that the received x~\tilde{x} has the same prediction as xx which is the reverse act w.r.t. the adversary. Hence, at the test time we use the certificate over the inverse ball B−1(x)={x~:x∈B(x~)}\mathcal{B}^{-1}(x) = \{\tilde{x} : x \in\mathcal{B}(\tilde{x})\} which is the smallest set that from any perturbed point x~∈B(x)\tilde{x} \in\mathcal{B}(x) it includes xx. This is the ball that covers the clean point from any potential perturbation. In symmetric balls, we have B−1(x)=B(x)\mathcal{B}^{-1}(x) = \mathcal{B}(x). An example for non-symmetric ball is token insertion, or deletion, where the inverse ball is the opposite act; i.e. Brins,−1=Brdel\mathcal{B}^{\mathrm{ins},-1}_r = \mathcal{B}^{\mathrm{del}}_r.

Binary and confidence certificate. We use c↓c^{\downarrow} and c↑c^{\uparrow} for certified lower and upper bounds, with a slight overload: formally c↓(f,B(x))≤inf⁡x~∈B(x)f(x~)c^{\downarrow}(f, \mathcal{B}(x)) \le\inf_{\tilde{x} \in\mathcal{B}(x)} f(\tilde{x}) and c↑c^{\uparrow} is the corresponding upper bound defined with supremum. In case that the bounds only depend on the value p=f(x)p = f(x) (not the function, and the point), we write c↓(p,B)c^{\downarrow}(p, \mathcal{B}) and c↑(p,B)c^{\uparrow}(p, \mathcal{B}). Certificates often report c↓c^{\downarrow}, and c↑c^{\uparrow}, from which we obtain the 0/1 decision or the certified radii – the largest radius for which the prediction is provably invariant; i.e. rx∗=max⁡{r:c↓(f,Br(x))≥1/2}r_x^* = \max\{r : c^{\downarrow}(f, \mathcal{B}_r(x)) \ge1/2\}. The latter is a common presentation of certificates; larger radii indicate stronger robustness. However ultimately the certification of a point boils down to comparison with a predefined radius rr, which is a binary decision. We say xx is certified at radius rr either if c↓(py∗(x),Br)>1/2c^{\downarrow}(p_y^*(x), \mathcal{B}_r) > 1/2 or if c↑(py(x),Br)>max⁡y≠y∗c↑(py(x),Br)c^{\uparrow}(p_y(x), \mathcal{B}_r) > \max_{y \ne y^*} c^{\uparrow}(p_y(x), \mathcal{B}_r), for y∗y^* as the top predicted class. In this work, we operate directly on these certified bounds.

Common certificates are often obtained via exact verification methods [38, 24] and randomized smoothing [11, 40]. The former is a white-box method that requires access to the model’s architecture and parameters, and is typically limited to smaller models of known architectures. The latter is a black-box method that can be applied to any model, and can scale to large and modern architectures, however it requires a large number of model forward passes to estimate the smoothed class probabilities with high confidence.

Objective. We replace the sound but expensive certificate with a tuned proxy function that controls the marginal false-approval rate at level α\alpha. The guarantee holds for all adversarial perturbations x~n+1∈B(xn+1)\tilde{x}_{n+1} \in\mathcal{B}(x_{n+1}) for any given clean point xn+1x_{n+1}, and the expectation in Eq. (2) is taken marginally over the distribution of (latent) clean inputs. We assume a calibration dataset Dn={x1,…,xn}\mathcal{D}_n = \{x_1, \ldots, x_n\} of unlabeled points drawn exchangeably from the same distribution as the clean test input xn+1x_{n+1} over which we tune the certificate. Only the latent clean point xn+1x_{n+1}, and not x~n+1\tilde{x}_{n+1}, needs to be exchangeable with Dn\mathcal{D}_n; e.g. for unseen deployment inputs, Dn\mathcal{D}_n must be held out from the training data of ff, as training and test points are not exchangeable conditional on the model. The outcome is a proxy that takes significantly lower computational cost only by relaxing the soundness of the original certificate to a marginally bounded false-approval rate; formally with cert⁡f,r\operatorname{cert}_{f,r}, we guarantee that

ED[{f(x~n+1)≠f(xn+1)}⋅cert⁡f,r(x~n+1)∣x~n+1∈B(xn+1)]≤α.(2)\mathbb{E}_{\mathcal{D}}\left[\left\{f(\tilde{x}_{n+1}) \ne f(x_{n+1})\right\} \cdot\operatorname{cert}_{f,r}(\tilde{x}_{n+1}) \mid\tilde{x}_{n+1} \in\mathcal{B}(x_{n+1})\right] \le\alpha. \tag*{(2)}

In words, over many future (possibly perturbed) inputs, on average at most an α\alpha fraction are approved while their prediction is flipped. The expectation is over the calibration set, the clean test point, and the internal randomness. Same as conformal prediction itself, the risk is bounded in expectation, and for a fixed calibration set can exceed α\alpha (with a variance decreasing in the number of calibration points nn); the guarantee is also not conditional on any particular input, class, or subpopulation.

Our proxy certificate (ProxC) is built upon two key components, which we introduce next.

Randomized Smoothing

We recall a high level understanding of randomized smoothing, and defer the details to § B. A smoothing function ψ:X→X\psi:\mathcal{X}\to\mathcal{X} augments the input with a random noise from a predefined distribution; ψ(x)=x+ϵ\psi(x)=x+\epsilon. Wrapping any base classifier ff, we define the smoothed class probability g(x)y=Pr⁡ϵ[f(x+ϵ)=y]g(x)_y=\Pr_{\epsilon}[f(x+\epsilon)=y]. Because the push-forward measures f#ψ(x)f_{\#}\psi(x) and f#ψ(x~)f_{\#}\psi(\widetilde{x}) overlap heavily for nearby inputs, we can derive model-agnostic lower and upper bounds on g(x~)yg(\widetilde{x})_y that depend only on p=g(x)yp=g(x)_y and the perturbation radius. On images a well-known smoothing scheme is the isotropic Gaussian noise of magnitude σ\sigma, i.e. ϵ∼N(0,σ2I)\epsilon\sim\mathcal{N}(0,\sigma^2I). For ℓ1\ell_1 and ℓ2\ell_2 ball, isotropic Gaussian smoothing has closed form lower and upper bounds (c↓c^{\downarrow} and c↑c^{\uparrow}) as Φσ(Φ−1(p)±r)\Phi_\sigma(\Phi^{-1}(p)\pm r) regardless of the model’s architecture, or the actual value of xx. Here Φσ\Phi_\sigma is the CDF of N(0,σ2I)\mathcal{N}(0,\sigma^2I). For other smoothing schemes, similar bounds can be computed through the recipe from Yang et al. [ [40].

Conformal Risk Control

References

  1. [1]Anastasios N Angelopoulos, Stephen Bates, Adam Fisch, Lihua Lei, and Tal Schuster. Conformal risk control. arXiv preprint arXiv:2208.02814, 2022.arxiv.org/abs/2208.02814
  2. [2]Anastasios N Angelopoulos, Rina Foygel Barber, and Stephen Bates. Theoretical foundations of conformal prediction. arXiv preprint arXiv:2411.11824, 2024.arxiv.org/abs/2411.11824
  3. [3]Anastasios N Angelopoulos, Stephen Bates, Emmanuel J Candès, Michael I Jordan, and Lihua Lei. Learn then test: Calibrating predictive algorithms to achieve risk control. The Annals of Applied Statistics, 19(2):1641–1662, 2025.arxiv.org/abs/2110.01052
  4. [4]Anastasios Nikolas Angelopoulos and Stephen Bates. A gentle introduction to conformal prediction and distribution-free uncertainty quantification. arXiv, abs/2107.07511, 2021. URL https://api.semanticscholar.org/CorpusID:235899036.arxiv.org/abs/2107.07511
  5. [5]Anish Athalye, Nicholas Carlini, and David Wagner. Obfuscated gradients give a false sense of security: Circumventing defenses to adversarial examples. In International conference on machine learning, pages 274–283. PMLR, 2018.arxiv.org/abs/1802.00420
  6. [7]Rina Foygel Barber, Emmanuel J. Candès, Aaditya Ramdas, and Ryan J. Tibshirani. Conformal prediction beyond exchangeability. The Annals of Statistics, 2022. URL https://api.semanticscholar.org/CorpusID:247158820.arxiv.org/abs/2202.13415
  7. [8]Stephen Bates, Anastasios Angelopoulos, Lihua Lei, Jitendra Malik, and Michael I. Jordan. Distribution-free, risk-controlling prediction sets. Journal of the ACM, 68(6):1–34, 2021.arxiv.org/abs/2101.02703
  8. [9]Peter Blohm, Patrick Indri, Thomas Gärtner, and Sagar Malhotra. Probably approximately global robustness certification. arXiv preprint arXiv:2511.06495, 2025.arxiv.org/abs/2511.06495
  9. [10]Nicholas Carlini, Florian Tramer, Krishnamurthy Dj Dvijotham, Leslie Rice, Mingjie Sun, and J Zico Kolter. (certified!!) adversarial robustness for free! arXiv preprint arXiv:2206.10550, 2022.arxiv.org/abs/2206.10550
  10. [11]Jeremy Cohen, Elan Rosenfeld, and Zico Kolter. Certified adversarial robustness via randomized smoothing. In international conference on machine learning, pages 1310–1320. PMLR, 2019.arxiv.org/abs/1902.02918
  11. [12]Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee, 2009.
  12. [13]Marc Fischer, Maximilian Baader, and Martin Vechev. Scalable certified segmentation via randomized smoothing. In International Conference on Machine Learning, pages 3340–3351. PMLR, 2021.arxiv.org/abs/2107.00228
  13. [14]Yonatan Geifman and Ran El-Yaniv. Selective classification for deep neural networks. In Advances in Neural Information Processing Systems, volume 30, pages 4885–4894, 2017.
  14. [15]Yonatan Geifman and Ran El-Yaniv. Selectivenet: A deep neural network with an integrated reject option. In Proceedings of the 36th International Conference on Machine Learning, volume 97 of Proceedings of Machine Learning Research, pages 2151–2159. PMLR, 2019.arxiv.org/abs/1901.09192
  15. [16]Asaf Gendler, Tsui-Wei Weng, Luca Daniel, and Yaniv Romano. Adversarially robust conformal prediction. In International Conference on Learning Representations, 2021.
  16. [17]Subhankar Ghosh, Yuanjie Shi, Taha Belkhouja, Yan Yan, Janardhan Rao Doppa, and Brian Jones. Probabilistically robust conformal prediction. In Conference on Uncertainty in Artificial Intelligence, 2023. URL https://api.semanticscholar.org/CorpusID:260334753.arxiv.org/abs/2307.16360
  17. [18]Ian J Goodfellow, Jonathon Shlens, and Christian Szegedy. Explaining and harnessing adversarial examples. arXiv preprint arXiv:1412.6572, 2014.DOI
  18. [19]Kavya Gupta and Sagar Verma. CertViT: Certified robustness of pre-trained vision transformers. arXiv preprint arXiv:2302.10287, 2023. URL https://arxiv.org/abs/2302.10287.
  19. [20]Kai Hu, Andy Zou, Zifan Wang, Klas Leino, and Matt Fredrikson. Unlocking deterministic robustness certification on imagenet. Advances in Neural Information Processing Systems, 36:42993–43011, 2023.arxiv.org/abs/2301.12549
  20. [21]Kai Hu, Haoqi Hu, and Matt Fredrikson. LipNeXt: Scaling up Lipschitz-based certified robustness to billion-parameter models. In International Conference on Learning Representations (ICLR), 2026. URL https://openreview.net/forum?id=2fgzEJPHO2.arxiv.org/abs/2601.18513
  21. [22]Zhuoqun Huang, Neil G Marchant, Olga O Obrimenko, and Benjamin IP Rubinstein. Cert-ed: Certifiably robust text classification for edit distance. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 10813–10835, 2024.arxiv.org/abs/2408.00728
  22. [23]Linus Jeary, Tom Kuipers, Mehran Hosseini, and Nicola Paoletti. Verifiably robust conformal prediction. 2024.arxiv.org/abs/2405.18942
  23. [24]Guy Katz, Clark Barrett, David L. Dill, Kyle Julian, and Mykel J. Kochenderfer. Reluplex: An efficient smt solver for verifying deep neural networks. In Computer Aided Verification, pages 97–117. Springer, 2017.
  24. [25]Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny images. 2009.
  25. [26]Aounon Kumar, Alexander Levine, Soheil Feizi, and Tom Goldstein. Certifying confidence via randomized smoothing. Advances in Neural Information Processing Systems, 33:5165–5177, 2020.arxiv.org/abs/2009.08061
  26. [27]Guang-He Lee, Yang Yuan, Shiyu Chang, and Tommi Jaakkola. Tight certificates of adversarial robustness for randomly smoothed classifiers. Advances in Neural Information Processing Systems, 32, 2019.
  27. [28]Thomas Massena, Léo Andéol, Thibaut Boissin, Corentin Friedrich, Franck Mamalet, Mathieu Serrurier, and Sébastien Gherchinovitz. Efficient robust conformal prediction via lipschitz-bounded networks. 2025.
  28. [29]Leslie Rice, Anna Bair, Huan Zhang, and J Zico Kolter. Robustness between the worst and average case. Advances in Neural Information Processing Systems, 34:27840–27851, 2021.
  29. [30]Yaniv Romano, Matteo Sesia, and Emmanuel J. Candès. Classification with valid and adaptive coverage. arXiv: Methodology, 2020.arxiv.org/abs/2006.02544
  30. [32]Hadi Salman, Jerry Li, Ilya Razenshteyn, Pengchuan Zhang, Huan Zhang, Sebastien Bubeck, and Greg Yang. Provably robust deep learning via adversarially trained smoothed classifiers. Advances in neural information processing systems, 32, 2019.
  31. [33]Hadi Salman, Mingjie Sun, Greg Yang, Ashish Kapoor, and J Zico Kolter. Denoised smoothing: A provable defense for pretrained classifiers. Advances in Neural Information Processing Systems, 33:21945–21957, 2020.arxiv.org/abs/2003.01908
  32. [34]Vincent Tjeng, Kai Y. Xiao, and Russ Tedrake. Evaluating robustness of neural networks with mixed integer programming. In International Conference on Learning Representations, 2019.
  33. [35]Václav Voráček. Treatment of statistical estimation problems in randomized smoothing for adversarial robustness. In Advances in Neural Information Processing Systems, volume 37, pages 133464–133486, 2024.
  34. [36]Vladimir Vovk, Alexander Gammerman, and Glenn Shafer. Algorithmic learning in a random world. 2005.
  35. [37]Lily Weng, Huan Zhang, Hongge Chen, Zhao Song, Cho-Jui Hsieh, Luca Daniel, Duane Boning, and Inderjit Dhillon. Towards fast computation of certified robustness for relu networks. In Proceedings of the 35th International Conference on Machine Learning, volume 80, pages 5276–5285. PMLR, 2018.arxiv.org/abs/1804.09699
  36. [38]Eric Wong and J. Zico Kolter. Provable defenses against adversarial examples via the convex outer adversarial polytope. In Proceedings of the 35th International Conference on Machine Learning, volume 80, pages 5286–5295. PMLR, 2018.
  37. [39]Ge Yan, Yaniv Romano, and Tsui-Wei Weng. Provably robust conformal prediction with improved efficiency. arXiv preprint arXiv:2404.19651, 2024.arxiv.org/abs/2404.19651
  38. [40]Greg Yang, Tony Duan, J. Edward Hu, Hadi Salman, Ilya Razenshteyn, and Jerry Li. Randomized smoothing of all shapes and sizes. In Proceedings of the 37th International Conference on Machine Learning, volume 119 of Proceedings of Machine Learning Research, pages 10693–10705. PMLR, 2020.arxiv.org/abs/2002.08118
  39. [41]Soroush H Zargarbashi and Aleksandar Bojchevski. Robust conformal prediction with a single binary certificate. 2025.
  40. [42]Soroush H Zargarbashi, Mohammad Sadegh Akhondzadeh, and Aleksandar Bojchevski. Robust yet efficient conformal prediction sets. In Forty-first International Conference on Machine Learning, 2024.
  41. [44]Huan Zhang, Tsui-Wei Weng, Pin-Yu Chen, Cho-Jui Hsieh, and Luca Daniel. Efficient neural network robustness certification with general activation functions. In Advances in Neural Information Processing Systems, volume 31, 2018.arxiv.org/abs/1811.00866

Paper details

vlnc:W_3f7qyerz
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 17

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Not recorded for this paper

View statement
Looking for related papers…
Contents