Pith. sign in

REVIEW 3 major objections 5 minor 68 references

Predictive Inference With Fast Feature Conformal Prediction

T0 review · 3 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 A Taylor-expanded conformity score makes Feature Conformal Prediction 50x faster while preserving coverage and band length

desk verdict Useful gradient-normalized score with a sound fixed-layer coverage guarantee, but the headline band-length win is an oracle over layers and the efficiency theorem essentially assumes the desired conclusion. read the letter →

arxiv 2412.00653 v1 pith:NLRWPU5S submitted 2024-12-01 cs.LG cs.AIstat.ML

Zihao Tang , Boyuan Wang , Chuan Wen , Jiaye Teng This is my paper · ORCID
classification cs.LGcs.AIstat.ML
keywords conformalpredictionfeaturespaceTaylorexpansionnon-conformityscoreuncertaintyquantificationbandlengthgradientnormalizationintervals
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

FFCP is a conformal-prediction method that takes Feature Conformal Prediction (FCP)—which builds prediction bands in a neural network's feature space instead of the output space—and replaces the expensive nonlinear step that maps bands back to output space with a first-order Taylor expansion of the prediction head. The resulting non-conformity score is a simple ratio: the absolute prediction error divided by the norm of the gradient of the head at the feature. The paper claims, and verifies experimentally, that this score is a faithful fast analogue of FCP: coverage remains at the target level, band lengths are comparable to FCP's and often shorter than vanilla conformal prediction's, and runtime drops by roughly 50x on the tested regression datasets. The practical point is that feature-space conformal prediction becomes cheap enough for routine use, and the gradient-normalization idea transfers to quantile regression, localized conformal prediction, classification, and segmentation.

What carries the argument

The load-bearing object is the gradient-normalized non-conformity score $s_{\mathrm{ff}}(X,Y,g\circ h)=|Y-f(X)|/\|\nabla g(\hat v)\|$, where $\hat v=h(X)$ is the feature embedding and $\nabla g(\hat v)$ is the Jacobian of the prediction head evaluated at that feature. It is the exact expression FCP's feature-space score reduces to when the head is linearized, and it makes the feature-to-output band mapping a single multiplication by $\|\nabla g(\hat v)\|$ instead of the expensive LiPRA optimization. The paper's efficiency theorem rests on the square conditions: expansion, meaning the feature space stretches the gap between individual scores and their quantile, and quantile stability, meaning the quantile computed on the calibration fold transfers to a fresh sample. Together these conditions make the quantile operation cheaper in feature space than in output space.

What would settle it

On any benchmark dataset, compute the mean absolute deviation of feature-space scores from their quantile, $M|Q_{1-\alpha}(V^o_D/\|\nabla g(\hat v)\|)-V^o_D/\|\nabla g(\hat v)\||$, and compare it with the corresponding output-space quantity $M[Q_{1-\alpha}(V^o_D)-V^o_D]$; if the feature-space quantity is not smaller, Theorem 6's expansion condition fails. A direct check also appears in the paper's own Table 2, where on SYNTHETIC, STAR, and BIO the reported FFCP band length equals Vanilla CP's, so any claim that FFCP is universally shorter would be refuted on those datasets.

Watch

Extended reading notes

Core claim

The central claim is that FCP's two nonlinear operations—the feature-space distance $s_f(X,Y,g\circ h)=\inf_{v:g(v)=Y}\|v-\hat v\|$ and the band-estimation step $\{g(v):\|v-\hat v\|\le Q_{1-\alpha}\}$—can both be approximated by replacing $g$ with its first-order Taylor expansion around the feature $\hat v=h(X)$. This yields the score $s_{\mathrm{ff}}(X,Y,g\circ h)=|Y-f(X)|/\|\nabla g(\hat v)\|$ and the interval $[f(X)-\|\nabla g(\hat v)\|Q_{1-\alpha},\;f(X)+\|\nabla g(\hat v)\|Q_{1-\alpha}]$. Under the standard exchangeability assumption, the interval has marginal coverage at least $1-\alpha$ (Theorem 4). Under additional square conditions—expansion and quantile stability, meaning feature-space scores sit closer to their quantile than output-space scores—the average band is provably shorter than vanilla conformal prediction's (Theorem 6). The experiments report coverage above the nominal level on all datasets, band lengths comparable to FCP and mostly shorter than vanilla CP, and roughly 50x faster runtime than FCP.

Load-bearing premise

FFCP's guarantee of shorter bands assumes the square conditions: feature-space non-conformity scores must be closer to their quantile than output-space scores, a property the paper checks empirically on a single illustrative plot rather than on the benchmark datasets. Coverage remains valid without this assumption, but the claimed efficiency advantage over vanilla conformal prediction would not be proven.

Editorial extensions

If this is right

  • FFCP can be run on real regression problems where FCP's LiPRA-based band estimation is too slow, with the same distribution-free coverage guarantee.
  • The gradient-normalized score can replace the plain residual in other conformal frameworks, producing FFCQR, FFLCP, and FFRAPS that inherit the speed and coverage properties.
  • Because the method is layer-agnostic, practitioners can choose which feature layer to split at; the empirical bands vary by layer, and the shortest band is not always at the deepest layer.
  • On untrained networks FFCP degrades to about vanilla CP length, so the efficiency gain is tied to the quality of learned feature representations.
  • The band-length advantage is dataset-dependent: on some datasets (e.g., SYNTHETIC, STAR, BIO) the reported lengths match vanilla CP exactly rather than improving.

Reading between the lines

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

  • The square conditions are verified only in one illustrative plot (Figure 4), not on the benchmark tables; if they fail, Theorem 6's band-length conclusion does not apply, even though Theorem 4's coverage still holds.
  • The paper's own untrained-network experiment and its closing remark about zero-gradient instability identify the main failure mode: when the gradient norm is uninformative or near zero, the normalized score inherits that noise, so the speedup comes with a representation-quality caveat.
  • The 50x speedup is reported against FCP's LiPRA implementation; on very high-dimensional outputs the gradient computation itself may dominate, so the practical speedup could shrink where the Jacobian is expensive.
  • A natural extension is to include second-order Taylor terms or a curvature correction and test whether intervals shorten further without breaking coverage, which the paper leaves as future work.
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

3 major / 5 minor

Summary. The paper proposes Fast Feature Conformal Prediction (FFCP), a conformal prediction method that replaces FCP's expensive feature-to-output band transformation with a first-order Taylor approximation. The resulting non-conformity score is |Y - f(X)| / ||∇g(h(X))||, and the prediction band is [f(X) - ||∇g(h(X))|| Q, f(X) + ||∇g(h(X))|| Q] for a calibration quantile Q. The authors prove finite-sample coverage under exchangeability (Theorem 4), claim a band-length advantage over Vanilla CP under 'square conditions' (Theorem 5/6), and report experiments on regression, classification, and segmentation showing roughly 50x speedup over FCP and shorter or equal band lengths relative to Vanilla CP. Extensions to CQR, LCP, and RAPS are also presented.

Significance. If the efficiency claim is validated, the paper is a useful contribution: it offers a simple, computationally cheap score function that preserves the split-conformal coverage guarantee while avoiding LiPRA's expensive nonlinear band estimation. The coverage argument in Theorem 4 is a standard and correct exchangeability argument, the Taylor derivation of the score is clean, and the runtime speedup over FCP in Table 1 is credible and is a genuine practical advantage. The code release and the extensions to CQR, LCP, and RAPS are also strengths. However, the central band-length claim currently rests on two unsupported pillars: post-hoc selection of the shortest layer on the test set, and Theorem 6's 'square conditions' that essentially assume the desired inequality. These issues affect the paper's headline claim, not just its presentation.

major comments (3)
  1. [Section 5.1 and Table 2] The reported FFCP band length is the shortest among the five network layers on the test set: Section 5.1 states that 'if only a single band length is presented, it corresponds to the shortest band length returned by the different neural network layers,' and the caption of Table 2 explicitly says 'we select the shortest band length among all layers.' However, Algorithm 2 defines FFCP for a fixed split f = g ∘ h and contains no layer-selection step. The evaluated predictor is therefore not the algorithm whose coverage is guaranteed by Theorem 4, and the exchangeability argument does not cover a data-dependent minimum of five score functions; a min-length selection rule can have marginal coverage below 1-α even when every fixed layer is marginally valid. The Table 2 band-length comparisons are thus oracle comparisons rather than comparisons of a single FFCP algorithm. Please report results for a pre-specified layer or for a selection rule defined before seeing test data, and state clearly whether any coverage statement applies to that rule.
  2. [Appendix A.2, Theorem 6] The 'Expansion' condition in Eq. (15) is L E_D~P^n |Q_{1-α}(V_o_D/||∇g(v)||) - V_o_D/||∇g(v)|||^α < E_D~P^n [Q_{1-α}(V_o_D) - V_o_D] - 2 max{L,1}(c/√n)^{min{α,1}}. This is essentially the inequality needed to conclude that the FFCP band is shorter in expectation, and the proof in Eq. (16)-(18) uses it directly. The 'Quantile Stability' assumption also compares quantiles across data sets with an unspecified constant c and an unproved O(1/√n) rate, and Theorem 6's statement says the feature space 'satisfies' these conditions while the proof assumes them. Thus the theorem is conditional on assumptions that are very close to the desired conclusion. Figure 4 checks the expansion only on one illustrative plot, at a layer that is itself selected for display, and not on the benchmark datasets. Please either prove the square conditions from primitive model/architecture assumptions, verify them quantitatively on all datasets, or explicitly state Theorem 6 as a conditional result with the conditions checked empirically.
  3. [Tables 2 and 5] The band-length comparison is further weakened by the construction of the min rule. Table 5 states that Layer 4 'is equivalent to Vanilla CP,' so the reported FFCP length, being a minimum over layers including Layer 4, is always no larger than the Vanilla CP length by construction. This makes the statement in Section 5.2 that 'FFCP surpasses Vanilla CP by achieving a shorter band length' impossible to interpret as evidence for the method's efficiency. Table 2 also shows equal lengths for SYNTHETIC, STAR, and BIO, so even the min rule does not uniformly improve on Vanilla. Please report the band lengths for each layer separately and compare a single fixed-layer FFCP with Vanilla CP.
minor comments (5)
  1. [Theorem 4 / Remark 3] The coordinate-wise extension in Eqs. (9)-(10) applies the conformal quantile separately to each coordinate, which provides coordinate-wise marginal coverage but not joint coverage of the full vector Y; the paper should state this limitation explicitly.
  2. [Section 5.2] The text says 'the coverage of FFCP all exceeds the confidence level 1 - α,' but Table 2 point estimates for FB2 (89.868), MEPS20 (89.615), and BIKE (89.624) are below 90%; the variability across runs may explain this, but the claim should be phrased as 'coverage is near or above the nominal level up to finite-sample variation.'
  3. [Appendix A.2] The notation in Theorem 6 is inconsistent: the output-space set is sometimes written V_o_D and sometimes V_o^D, the final inequality in Eq. (18) uses Q_{1-α}(V_o_D) instead of Q_{1-α}(V_o_Dcal), and the expectation over the test point is missing from the displayed conclusion. In addition, 'Holder' should be 'Hölder.'
  4. [Figure 4] The caption says 'FFCP selects layer 2 for display' but does not explain why layer 2 is chosen or how robust the visual check is across layers and datasets; a quantitative summary of Eq. (15)'s left and right sides would be more informative.
  5. [Various] There are several typos, including 'tesing point' in Algorithms 1, 2, and 4, 'meps19 detaset' in Section B.5, and 'discusses discusses' in Section B; these should be corrected.

Circularity Check

2 steps flagged · score 6.0 of 10

Efficiency claim is built into both the empirical metric (min over layers) and the theoretical premise (square conditions); coverage and speedup are independent.

  1. self definitional [Section 5.1 (band length evaluation) and Table 2 caption]
    "Since we use a 5-layer neural network, each layer can be viewed as a feature layer. Therefore, in the experiments, we obtain the band length returned by each of the 5 layers of the neural network. In the subsequent results, if only a single band length is presented, it corresponds to the shortest band length returned by the different neural network layers. ... For FFCP, we select the shortest band length among all layers."

    Algorithm 2 defines FFCP as a single split-point procedure: it trains f = g∘h, computes the score |Y−f(X)|/||∇g(ˆv)|| at one split, and returns one band; there is no layer-selection step. The empirical 'FFCP' band length in Table 2 is instead defined as the shortest band among five layers, with the last layer typically equal to Vanilla CP. The claim 'FFCP surpasses Vanilla CP by achieving a shorter band length' is therefore forced by the evaluation rule: the reported number is a minimum over five candidate intervals, not the output of Algorithm 2. Moreover, selecting the shortest interval after seeing the test set makes the selected interval a data-dependent predictor, so Theorem 4's exchangeability guarantee for a fixed score function does not apply to the reported coverage.

  2. other [Theorem 6 and Appendix A.2; informal statement in Theorem 5; verification in Appendix B.2]
    "Then the feature space satisfies the following square conditions: 1. Expansion. The feature space expands the differences between individual length and their quantiles, namely, L E_{D∼P^n} M|Q_{1−α}(V^o_D/∥∇g(ˆv)∥) − V^o_D/∥∇g(ˆv)∥|^α < E_{D∼P^n} M[Q_{1−α}(V^o_D) − V^o_D] − 2 max{L, 1}(c/√n)^{min{α,1}}. ... Then FFCP provably outperforms vanilla CP in terms of average band length, namely, E_{(X′,Y′)∼P}(∥∇g(ˆv′)∥ · Q_{1−α}(V^o_{Dcal}/∥∇g(ˆv_cal)∥) < Q_{1−α}(V^0_{Dcal})."

    The proof of Theorem 6 in Appendix A.2 begins by restating the Expansion assumption as the first inequality (labeled Eq. 15) and then, after Holder and Quantile Stability manipulations, concludes the target band-length inequality (Eq. 18). The Expansion premise is not derived from first principles; it directly asserts that the feature-space quantile gap is smaller than the output-space quantile gap, which is essentially the efficiency comparison that the theorem promises to prove. The premise is only 'validated' in Figure 4, a single illustrative plot, and Appendix B.2 states 'we take exponent α = 1 and do not consider the Lipschitz factor L,' so the condition is not checked on the benchmark datasets used for the efficiency claims.

full rationale

The coverage theorem (Theorem 4) is standard split-conformal exchangeability applied to the score |Y−f(X)|/||∇g(ˆv)|| and is not circular. The runtime speedup comparison (Table 1) is an engineering measurement after training and is credible. The central band-length advantage, however, is partially circular in two places. First, the reported FFCP band length in Table 2 is not the output of Algorithm 2, which has a fixed split point; it is the minimum over five layer-based intervals, with the last layer typically equal to Vanilla CP. Choosing the shortest interval after seeing the test set makes 'FFCP outperforms Vanilla CP' true by construction of the reported metric, and the coverage of that selected interval is not covered by Theorem 4 because the selected score is not a fixed function of (X,Y). Second, the formal efficiency theorem assumes square conditions whose Expansion inequality is, up to Holder and stability slack terms, the same feature-vs-output quantile-gap comparison that the theorem concludes; the Appendix A.2 proof starts from this assumption and rearranges it into the desired band-length inequality. The assumption is only illustrated in Figure 4, not tested on the benchmark datasets. These are partial circularities in the central efficiency claim. A fixed-layer evaluation with square conditions verified per dataset would restore a non-circular efficiency test. The self-citation to Teng et al. (2022) for FCP is not load-bearing here: the coverage argument is self-contained, and the runtime speedup does not depend on the FCP efficiency theorem.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The central method introduces no new physical or mathematical entities. The main load-bearing assumptions are exchangeability (standard), the accuracy of the Taylor linearization, the nonzero gradient norm, and the square conditions for the efficiency theorem. The only fitted quantity in the evaluation is the layer selection, which is chosen post hoc on the test set.

free parameters (2)
  • Layer selection for FFCP band length = best of layers 0 to 4 on each dataset's test set
    Reported FFCP band length is the minimum over five network layers (Section 5.1, Table 2), a data-dependent choice that can inflate apparent performance and includes a free selection step in the evaluation.
  • FFRAPS regularization hyperparameter delta = not reported in main text
    Algorithm 5 uses an adjustable hyperparameter delta for the classification extension; it is tuned per dataset and not part of the central FFCP method.
assumptions (5)
  • domain assumption Assumption 1: calibration and test points are exchangeable given the trained model
    Standard conformal prediction assumption (Section 3), required for the coverage guarantee in Theorem 4.
  • ad hoc to paper The prediction head g is approximately linear over the feature-space ball of radius Q (first-order Taylor expansion)
    Used in Section 4.1 to derive the FFCP score and the output band. No bound on approximation error is provided; if the head is highly nonlinear, FFCP band lengths may diverge from FCP's.
  • domain assumption The gradient norm ||grad g(v_hat)|| is nonzero and stable
    FFCP divides by the gradient norm (Equation 2). The paper lists zero gradients as a limitation in the conclusion, but the main method has no guard.
  • ad hoc to paper Square conditions: Expansion and Quantile Stability (Theorem 6)
    These conditions are introduced to prove the band length advantage and are essentially the conclusion restated; they are only empirically validated on one plot in Appendix B.2.
  • standard math Holder/Lipschitz condition on quantile-related functions (Theorem 6)
    Used in Appendix A.2 to derive the band length inequality; a regularity condition assumed without empirical support.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Predictive Inference With Fast Feature Conformal Prediction." pith.science (2026). https://pith.science/paper/NLRWPU5S

@misc{pith2026241200653,
  author       = {Pith},
  title        = {Pith review of: Predictive Inference With Fast Feature Conformal Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NLRWPU5S}},
  note         = {Machine review of arXiv:2412.00653}
}
read the original abstract

Conformal prediction is widely adopted in uncertainty quantification, due to its post-hoc, distribution-free, and model-agnostic properties. In the realm of modern deep learning, researchers have proposed Feature Conformal Prediction (FCP), which deploys conformal prediction in a feature space, yielding reduced band lengths. However, the practical utility of FCP is limited due to the time-consuming non-linear operations required to transform confidence bands from feature space to output space. In this paper, we introduce Fast Feature Conformal Prediction (FFCP), which features a novel non-conformity score and is convenient for practical applications. FFCP serves as a fast version of FCP, in that it equivalently employs a Taylor expansion to approximate the aforementioned non-linear operations in FCP. Empirical validations showcase that FFCP performs comparably with FCP (both outperforming the vanilla version) while achieving a significant reduction in computational time by approximately 50x. The code is available at https://github.com/ElvisWang1111/FastFeatureCP

Figures

Figures reproduced from arXiv: 2412.00653 by the authors.

Figure 1
Figure 1. Comparison among Vanilla CP, FCP, and FFCP. FCP and FFCP are more efficient [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The results of FFCP and FCP in image segmentation tasks show that brighter [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗
Figure 3
Figure 3. Scatter plot of FCP Score and FFCP Score at different Layers. The relationship [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Empirical validation of Theorem 6. We plot the score distributions and their [PITH_FULL_IMAGE:figures/full_fig_p027_4.png]
Figure 5
Figure 5. Figure 5: Additional visualization results in segmentation task. [PITH_FULL_IMAGE:figures/full_fig_p033_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

68 extracted references · 36 canonical work pages

  1. [1]

    Machine learning: Trends, perspectives, and prospects

    Michael I Jordan and Tom M Mitchell. Machine learning: Trends, perspectives, and prospects. Science, 349 0 (6245): 0 255--260, 2015

    2015

  2. [2]

    Mastering the game of go without human knowledge

    David Silver, Julian Schrittwieser, Karen Simonyan, Ioannis Antonoglou, Aja Huang, Arthur Guez, Thomas Hubert, Lucas Baker, Matthew Lai, Adrian Bolton, et al. Mastering the game of go without human knowledge. nature, 550 0 (7676): 0 354--359, 2017

    2017

  3. [3]

    Mitigating neural network overconfidence with logit normalization

    Hongxin Wei, Renchunzi Xie, Hao Cheng, Lei Feng, Bo An, and Yixuan Li. Mitigating neural network overconfidence with logit normalization. In International conference on machine learning, pages 23631--23644. PMLR, 2022

    work page 2022

  4. [4]

    Survey of hallucination in natural language generation

    Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Ye Jin Bang, Andrea Madotto, and Pascale Fung. Survey of hallucination in natural language generation. ACM Computing Surveys, 55 0 (12): 0 1--38, 2023

    2023

  5. [5]

    Uncertainty and technological change in medicine

    Annetine C Gelijns, Joshua Graff Zivin, and Richard R Nelson. Uncertainty and technological change in medicine. Journal of Health Politics, Policy and Law, 26 0 (5): 0 913--924, 2001

    work page 2001

  6. [6]

    The uncertain effect of financial incentives to improve health behaviors

    Harsha Thirumurthy, David A Asch, and Kevin G Volpp. The uncertain effect of financial incentives to improve health behaviors. Jama, 321 0 (15): 0 1451--1452, 2019

    work page 2019

  7. [7]

    The financial diaries: How American families cope in a world of uncertainty

    Jonathan Morduch and Rachel Schneider. The financial diaries: How American families cope in a world of uncertainty. Princeton University Press, 2017

    work page 2017

  8. [8]

    A review of uncertainty quantification in deep learning: Techniques, applications and challenges

    Moloud Abdar, Farhad Pourpanah, Sadiq Hussain, Dana Rezazadegan, Li Liu, Mohammad Ghavamzadeh, Paul Fieguth, Xiaochun Cao, Abbas Khosravi, U Rajendra Acharya, et al. A review of uncertainty quantification in deep learning: Techniques, applications and challenges. Information fusion, 76: 0 243--297, 2021

    2021

Show all 68 references
  1. [9]

    Weinberger

    Chuan Guo, Geoff Pleiss, Yu Sun, and Kilian Q. Weinberger. On calibration of modern neural networks. ArXiv, abs/1706.04599, 2017

    2017 arXiv

  2. [10]

    Gutmann, Aaron Courville, and Zhanxing Zhu

    Yanzhi Chen, Dinghuai Zhang, Michael U. Gutmann, Aaron Courville, and Zhanxing Zhu. Neural approximate sufficient statistics for implicit models. In International Conference on Learning Representations, 2021. URL https://openreview.net/forum?id=SRDuJssQud

    2021

  3. [11]

    Kruspe, Rudolph Triebel, Peter Jung, Ribana Roscher, Muhammad Shahzad, Wen Yang, Richard Bamler, and Xiao Xiang Zhu

    Jakob Gawlikowski, Cedrique Rovile Njieutcheu Tassi, Mohsin Ali, Jongseok Lee, Matthias Humt, Jianxiang Feng, Anna M. Kruspe, Rudolph Triebel, Peter Jung, Ribana Roscher, Muhammad Shahzad, Wen Yang, Richard Bamler, and Xiao Xiang Zhu. A survey of uncertainty in deep neural net...

    2021 arXiv

  4. [12]

    Algorithmic learning in a random world, volume 29

    Vladimir Vovk, Alexander Gammerman, and Glenn Shafer. Algorithmic learning in a random world, volume 29. Springer, 2005

    2005

  5. [13]

    A tutorial on conformal prediction

    Glenn Shafer and Vladimir Vovk. A tutorial on conformal prediction. Journal of Machine Learning Research, 9 0 (3), 2008 a

    2008

  6. [14]

    Efficiency of conformalized ridge regression

    Evgeny Burnaev and Vladimir Vovk. Efficiency of conformalized ridge regression. In Conference on Learning Theory, pages 605--622. PMLR, 2014

    2014

  7. [15]

    Learning semantic representations using convolutional neural networks for web search

    Yelong Shen, Xiaodong He, Jianfeng Gao, Li Deng, and Gr \'e goire Mesnil. Learning semantic representations using convolutional neural networks for web search. In Proceedings of the 23rd international conference on world wide web, pages 373--374, 2014

    2014

  8. [16]

    Predictive inference with feature conformal prediction

    Jiaye Teng, Chuan Wen, Dinghuai Zhang, Yoshua Bengio, Yang Gao, and Yang Yuan. Predictive inference with feature conformal prediction. arXiv preprint arXiv:2210.00173, 2022

    2022 arXiv

  9. [17]

    Automatic perturbation analysis for scalable certified robustness and beyond

    Kaidi Xu, Zhouxing Shi, Huan Zhang, Yihan Wang, Kai-Wei Chang, Minlie Huang, Bhavya Kailkhura, Xue Lin, and Cho-Jui Hsieh. Automatic perturbation analysis for scalable certified robustness and beyond. Advances in Neural Information Processing Systems, 33: 0 1129--1141, 2020

    2020

  10. [18]

    Conformalized quantile regression

    Yaniv Romano, Evan Patterson, and Emmanuel Candes. Conformalized quantile regression. Advances in neural information processing systems, 32, 2019 a

    2019

  11. [19]

    Localized conformal prediction: A generalized inference framework for conformal prediction

    Leying Guan. Localized conformal prediction: A generalized inference framework for conformal prediction. Biometrika, 110 0 (1): 0 33--50, 2023

    2023

  12. [20]

    Uncertainty sets for image classifiers using conformal prediction

    Anastasios Angelopoulos, Stephen Bates, Jitendra Malik, and Michael I Jordan. Uncertainty sets for image classifiers using conformal prediction. arXiv preprint arXiv:2009.14193, 2020

    2009 arXiv

  13. [21]

    A tutorial on conformal prediction

    Glenn Shafer and Vladimir Vovk. A tutorial on conformal prediction. J. Mach. Learn. Res., 9: 0 371--421, 2008 b . URL https://dl.acm.org/citation.cfm?id=1390693

    2008

  14. [22]

    Cand \`e s, Aaditya Ramdas, and Ryan J

    Rina Foygel Barber, Emmanuel J. Cand \`e s, Aaditya Ramdas, and Ryan J. Tibshirani. The limits of distribution-free conditional predictive inference. Information and Inference: A Journal of the IMA, 2020

    2020

  15. [23]

    Benchmarking llms via uncertainty quantification

    Fanghua Ye, Mingming Yang, Jianhui Pang, Longyue Wang, Derek F Wong, Emine Yilmaz, Shuming Shi, and Zhaopeng Tu. Benchmarking llms via uncertainty quantification. arXiv preprint arXiv:2401.12794, 2024

    2024 arXiv

  16. [24]

    Conformal prediction with large language models for multi-choice question answering

    Bhawesh Kumar, Charlie Lu, Gauri Gupta, Anil Palepu, David Bellamy, Ramesh Raskar, and Andrew Beam. Conformal prediction with large language models for multi-choice question answering. arXiv preprint arXiv:2305.18404, 2023

    2023 arXiv

  17. [25]

    Conformal language modeling

    Victor Quach, Adam Fisch, Tal Schuster, Adam Yala, Jae Ho Sohn, Tommi S Jaakkola, and Regina Barzilay. Conformal language modeling. arXiv preprint arXiv:2306.10193, 2023

    2023 arXiv

  18. [26]

    Tibshirani, Rina Foygel Barber, Emmanuel J

    Ryan J. Tibshirani, Rina Foygel Barber, Emmanuel J. Cand \` e s, and Aaditya Ramdas. Conformal prediction under covariate shift. In Hanna M. Wallach, Hugo Larochelle, Alina Beygelzimer, Florence d'Alch \' e - Buc, Emily B. Fox, and Roman Garnett, editors, Advances in Neural In...

    2019

  19. [27]

    A distribution-free test of covariate shift using conformal prediction

    Xiaoyu Hu and Jing Lei. A distribution-free test of covariate shift using conformal prediction. arXiv: Methodology, 2020

    2020

  20. [28]

    Distribution-free uncertainty quantification for classification under label shift

    Aleksandr Podkopaev and Aaditya Ramdas. Distribution-free uncertainty quantification for classification under label shift. In UAI, 2021

    2021

  21. [29]

    Conformal prediction beyond exchangeability

    Rina Foygel Barber, Emmanuel J Candes, Aaditya Ramdas, and Ryan J Tibshirani. Conformal prediction beyond exchangeability. arXiv preprint arXiv:2202.13415, 2022

    2022 arXiv

  22. [30]

    Cand \`e s

    Yaniv Romano, Matteo Sesia, and Emmanuel J. Cand \`e s. Classification with valid and adaptive coverage. arXiv: Methodology, 2020

    2020

  23. [31]

    Conformal prediction interval for dynamic time-series

    Chen Xu and Yao Xie. Conformal prediction interval for dynamic time-series. In ICML, 2021

    2021

  24. [32]

    Cand \`e s

    Isaac Gibbs and Emmanuel J. Cand \`e s. Adaptive conformal inference under distribution shift. In NeurIPS, 2021

    2021

  25. [33]

    T-SCI: A two-stage conformal inference algorithm with guaranteed coverage for cox-mlp

    Jiaye Teng, Zeren Tan, and Yang Yuan. T-SCI: A two-stage conformal inference algorithm with guaranteed coverage for cox-mlp. In Marina Meila and Tong Zhang, editors, Proceedings of the 38th International Conference on Machine Learning, ICML 2021, 18-24 July 2021, Virtual Event...

    2021

  26. [34]

    Conformalized survival analysis

    Emmanuel Cand \`e s, Lihua Lei, and Zhimei Ren. Conformalized survival analysis. Journal of the Royal Statistical Society Series B: Statistical Methodology, 85 0 (1): 0 24--45, 2023

    2023

  27. [35]

    Conformalized survival analysis

    Emmanuel J Cand \`e s, Lihua Lei, and Zhimei Ren. Conformalized survival analysis. arXiv preprint arXiv:2103.09763, 2021

    2021 arXiv

  28. [36]

    Wasserman

    Jing Lei, Alessandro Rinaldo, and Larry A. Wasserman. A conformal prediction approach to explore functional data. Annals of Mathematics and Artificial Intelligence, 74: 0 29--43, 2013

    2013

  29. [37]

    Bellman conformal inference: Calibrating prediction intervals for time series

    Zitong Yang, Emmanuel Cand \`e s, and Lihua Lei. Bellman conformal inference: Calibrating prediction intervals for time series. arXiv preprint arXiv:2402.05203, 2024

    2024 arXiv

  30. [38]

    Cand \`e s

    Lihua Lei and Emmanuel J. Cand \`e s. Conformal inference of counterfactuals and individual treatment effects. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 83, 2021

    2021

  31. [39]

    Regression conformal prediction with nearest neighbours

    Harris Papadopoulos, Vladimir Vovk, and Alexander Gammerman. Regression conformal prediction with nearest neighbours. J. Artif. Intell. Res., 40: 0 815--840, 2011 a . URL http://jair.org/papers/paper3198.html

    2011

  32. [40]

    Cand \` e s

    Yaniv Romano, Evan Patterson, and Emmanuel J. Cand \` e s. Conformalized quantile regression. In Hanna M. Wallach, Hugo Larochelle, Alina Beygelzimer, Florence d'Alch \' e - Buc, Emily B. Fox, and Roman Garnett, editors, Advances in Neural Information Processing Systems 32: An...

    2019

  33. [41]

    A comparison of some conformal quantile regression methods

    Matteo Sesia and Emmanuel J Cand \`e s. A comparison of some conformal quantile regression methods. Stat, 9 0 (1): 0 e261, 2020

    2020

  34. [42]

    Shimizu, and Rafael Bassi Stern

    Rafael Izbicki, Gilson T. Shimizu, and Rafael Bassi Stern. Distribution-free conditional predictive bands using density estimators. ArXiv, abs/1910.05575, 2020 a

    1910 arXiv

  35. [43]

    Conformal prediction using conditional histograms

    Matteo Sesia and Yaniv Romano. Conformal prediction using conditional histograms. In Marc'Aurelio Ranzato, Alina Beygelzimer, Yann N. Dauphin, Percy Liang, and Jennifer Wortman Vaughan, editors, Advances in Neural Information Processing Systems 34: Annual Conference on Neural ...

    2021

  36. [44]

    Cd-split and hpd-split: efficient conformal regions in high dimensions

    Rafael Izbicki, Gilson Shimizu, and Rafael B Stern. Cd-split and hpd-split: efficient conformal regions in high dimensions. arXiv preprint arXiv:2007.12778, 2020 b

    2007 arXiv

  37. [45]

    Finite-sample efficient conformal prediction

    Yachong Yang and Arun Kumar Kuchibhotla. Finite-sample efficient conformal prediction. arXiv preprint arXiv:2104.13871, 2021

    2021 arXiv

  38. [46]

    Split localized conformal prediction

    Xing Han, Ziyang Tang, Joydeep Ghosh, and Qiang Liu. Split localized conformal prediction. arXiv preprint arXiv:2206.13092, 2022

    2022 arXiv

  39. [47]

    Normalized nonconformity measures for regression conformal prediction

    Harris Papadopoulos, Alex Gammerman, and Volodya Vovk. Normalized nonconformity measures for regression conformal prediction. In Proceedings of the IASTED International Conference on Artificial Intelligence and Applications (AIA 2008), pages 64--69, 2008

    2008

  40. [48]

    Regression conformal prediction with nearest neighbours

    Harris Papadopoulos, Vladimir Vovk, and Alex Gammerman. Regression conformal prediction with nearest neighbours. Journal of Artificial Intelligence Research, 40: 0 815--840, 2011 b

    2011

  41. [49]

    Reliable prediction intervals with regression neural networks

    Harris Papadopoulos and Haris Haralambous. Reliable prediction intervals with regression neural networks. Neural Networks, 24 0 (8): 0 842--851, 2011

    2011

  42. [50]

    Improving adaptive conformal prediction using self-supervised learning

    Nabeel Seedat, Alan Jeffares, Fergus Imrie, and Mihaela van der Schaar. Improving adaptive conformal prediction using self-supervised learning. In International Conference on Artificial Intelligence and Statistics, pages 10160--10177. PMLR, 2023

    2023

  43. [51]

    Triage: Characterizing and auditing training data for improved regression

    Nabeel Seedat, Jonathan Crabb \'e , Zhaozhi Qian, and Mihaela van der Schaar. Triage: Characterizing and auditing training data for improved regression. Advances in Neural Information Processing Systems, 36, 2024

    2024

  44. [52]

    Accurate uncertainties for deep learning using calibrated regression

    Volodymyr Kuleshov, Nathan Fenner, and Stefano Ermon. Accurate uncertainties for deep learning using calibrated regression. In Jennifer G. Dy and Andreas Krause, editors, Proceedings of the 35th International Conference on Machine Learning, ICML 2018, Stockholmsm \" a ssan, St...

    2018

  45. [53]

    Dusenberry, Linchuan Zhang, Ghassen Jerfel, and Dustin Tran

    Jeremy Nixon, Michael W. Dusenberry, Linchuan Zhang, Ghassen Jerfel, and Dustin Tran. Measuring calibration in deep learning. In IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPR Workshops 2019, Long Beach, CA, USA, June 16-20, 2019 , pages 38--41. Com...

    2019

  46. [54]

    A survey on evaluation of large language models

    Yupeng Chang, Xu Wang, Jindong Wang, Yuan Wu, Linyi Yang, Kaijie Zhu, Hao Chen, Xiaoyuan Yi, Cunxiang Wang, Yidong Wang, et al. A survey on evaluation of large language models. ACM Transactions on Intelligent Systems and Technology, 15 0 (3): 0 1--45, 2024

    2024

  47. [55]

    Weight uncertainty in neural network

    Charles Blundell, Julien Cornebise, Koray Kavukcuoglu, and Daan Wierstra. Weight uncertainty in neural network. In Francis R. Bach and David M. Blei, editors, Proceedings of the 32nd International Conference on Machine Learning, ICML 2015, Lille, France, 6-11 July 2015 , volum...

    2015

  48. [56]

    Jos \' e Miguel Hern \' a ndez - Lobato and Ryan P. Adams. Probabilistic backpropagation for scalable learning of bayesian neural networks. In Francis R. Bach and David M. Blei, editors, Proceedings of the 32nd International Conference on Machine Learning, ICML 2015, Lille, Fr...

    2015

  49. [57]

    Dropout inference in bayesian neural networks with alpha-divergences

    Yingzhen Li and Yarin Gal. Dropout inference in bayesian neural networks with alpha-divergences. In Doina Precup and Yee Whye Teh, editors, Proceedings of the 34th International Conference on Machine Learning, ICML 2017, Sydney, NSW, Australia, 6-11 August 2017 , volume 70 of ...

    2017

  50. [58]

    What are bayesian neural network posteriors really like? In International conference on machine learning, pages 4629--4640

    Pavel Izmailov, Sharad Vikram, Matthew D Hoffman, and Andrew Gordon Gordon Wilson. What are bayesian neural network posteriors really like? In International conference on machine learning, pages 4629--4640. PMLR, 2021

    2021

  51. [59]

    Hands-on bayesian neural networks—a tutorial for deep learning users

    Laurent Valentin Jospin, Hamid Laga, Farid Boussaid, Wray Buntine, and Mohammed Bennamoun. Hands-on bayesian neural networks—a tutorial for deep learning users. IEEE Computational Intelligence Magazine, 17 0 (2): 0 29--48, 2022

    2022

  52. [60]

    Asuncion

    Arthur U. Asuncion. Uci machine learning repository, university of california, irvine, school of information and computer sciences. 2007

    2007

  53. [61]

    Cohen, Steven B

    Joel W. Cohen, Steven B. Cohen, and Jessica S. Banthin. The medical expenditure panel survey: A national information resource to support healthcare cost research and inform policy and practice. Medical Care, 47: 0 S44--S50, 2009

    2009

  54. [62]

    Tennessee’s student teacher achievement ratio (star) project

    CM Achilles, Helen Pate Bain, Fred Bellott, Jayne Boyd-Zaharias, Jeremy Finn, John Folger, John Johnston, and Elizabeth Word. Tennessee’s student teacher achievement ratio (star) project. Harvard Dataverse, 1: 0 2008, 2008

    2008

  55. [63]

    Feedback prediction for blogs

    Krisztian Buza. Feedback prediction for blogs. In Data analysis, machine learning and knowledge discovery, pages 145--152. Springer, 2014

    2014

  56. [64]

    The cityscapes dataset for semantic urban scene understanding

    Marius Cordts, Mohamed Omran, Sebastian Ramos, Timo Rehfeld, Markus Enzweiler, Rodrigo Benenson, Uwe Franke, Stefan Roth, and Bernt Schiele. The cityscapes dataset for semantic urban scene understanding. In Proc. of the IEEE Conference on Computer Vision and Pattern Recognitio...

    2016

  57. [65]

    Conformal prediction under covariate shift

    Ryan J Tibshirani, Rina Foygel Barber, Emmanuel Candes, and Aaditya Ramdas. Conformal prediction under covariate shift. Advances in neural information processing systems, 32, 2019 b

    2019

  58. [66]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should ...

  59. [67]

    \@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@firs...

  60. [68]

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibset...

    1940

Pith tools

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