Tabular Diffusion Counterfactual Explanations

Brian Barr; John Paisley; Wei Zhang

REVIEW 3 major objections 6 minor 44 references

Tabular Diffusion Counterfactual Explanations

T0 review · 3 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Tabular classifiers can be explained by guiding a diffusion reverse process, with categorical features relaxed through Gumbel-softmax so classifier gradients can flow.

desk verdict Useful empirical recipe for tabular counterfactuals via Gumbel-softmax guidance, but the theoretical bound in Theorem 4.1 is unproven—the proof rests on a false inequality. read the letter →

arxiv 2509.00876 v1 pith:WTRRHFU4 submitted 2025-08-31 cs.LG

Wei Zhang , Brian Barr , John Paisley This is my paper · ORCID

classification cs.LG

keywords counterfactualexplanationsdiffusionmodelstabulardataGumbel-softmaxclassifierguidancecategoricalfeaturesexplainablemachinelearningrecourse

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

Counterfactual explanations for tabular data have mostly been handled by latent-space searches or per-sample optimization; this paper proposes a diffusion-based alternative that works directly in the feature space. By relaxing categorical one-hot vectors through Gumbel-softmax, the reverse process can accept classifier gradients and guide the sample toward the target class without enumerating category combinations. The authors test on four credit and law datasets and report that their TDCE procedure achieves the highest validity, low JS divergence between counterfactual and target categorical distributions, and competitive interpretability, in a fixed number of reverse steps. A working version would give finance and social-science practitioners a model-agnostic explainer that does not require per-sample search.

What carries the argument

The load-bearing object is the Gumbel-softmax reparameterization of each categorical feature combined with a linearized guided reverse update. The relaxed vector replaces the one-hot value; the reverse log-density is approximated by the inner product of the relaxed vector with log probabilities from the denoiser, and the classifier's gradient is added to those logits. This makes the categorical reverse step resemble the Gaussian classifier-guidance update, allowing backpropagation through discrete features without enumerating all category combinations. The temperature controls how close the relaxation is to a one-hot vector and how much gradient variance appears.

What would settle it

Compute the guided and unguided reverse trajectories on the same datasets and compare validity and JS divergence; if they match, the categorical guidance is not the active mechanism. Also, the proof's key inequality (Appendix A, A.17) fails for simple draws like π=(0.95,0.05), x=(0.2,0.8), so the bound cannot be relied on as a guarantee.

Watch

Extended reading notes

Core claim

Central claim: a tabular diffusion model becomes a counterfactual explainer when classifier guidance is added to the reverse process and categorical features are relaxed with Gumbel-softmax vectors. Continuous features use the usual mean-plus-covariance gradient shift; categorical features approximate the Gumbel-softmax log-density linearly and add the classifier gradient to the logits (Eq. 18). The final category is an argmax. On four credit and law datasets, the paper reports the best validity, lower JS divergence for categorical distributions, and better interpretability than baselines at a fixed number of reverse steps, plus a temperature-dependent bound linking low temperature to a bett

Load-bearing premise

The categorical guidance pipeline assumes the approximation log pθ(ext|ext+1) ≈ ex⊤ log πθ(ext+1) + const is accurate enough that the guided reverse process still produces realistic categorical distributions; the paper's stated bound for this closeness relies on an inequality that does not hold for all Gumbel-softmax draws, so the theoretical guarantee is not actually established.

Editorial extensions

If this is right

Counterfactual generation for categorical features becomes a fixed-cost reverse pass rather than a per-sample search, so explanation time scales with the diffusion depth, not with the data size.
Continuous and categorical features are handled jointly in the same reverse process, so explanations can respect numeric and discrete structure at the same time.
Immutable features can be enforced with a mask that blends the noisy input into the guided sample, which keeps protected or unchangeable attributes fixed.
Temperature becomes a practical dial: lower temperatures make the relaxed features closer to one-hot but increase gradient variance, so each dataset needs its own temperature setting.

Reading between the lines

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

Because Eq. (16) linearizes the Gumbel-softmax log-density, the guidance update effectively shifts logits by a constant gradient; a natural extension is to learn a per-step logit correction that keeps the reverse trajectory on the data manifold even at high temperature.
The approximation error is controlled by the minimum relaxed value, so categorical variables with skewed category distributions are the likely failure mode; testing on datasets with many rare categories would show where the method's validity advantage erodes.
If the reported validity and interpretability gains hold, they imply that feature-space diffusion explainers stay closer to the data manifold than latent-space VAE explainers; this is directly testable by measuring the distance from generated counterfactuals to a held-out training set.

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.

Desk Editor's Note

read the letter

Mike,

Quick take on arXiv:2509.00876. The paper proposes TDCE, a diffusion-based counterfactual generator for tabular data that handles continuous and categorical features together. The key trick is to relax categorical one-hot vectors with Gumbel-softmax during the reverse diffusion, so you can backprop classifier gradients through them and do guidance similar to the Gaussian case. That is a sensible extension of Schiff et al. and Gruver et al. to the tabular mixed-type setting, and the empirical results on four datasets look decent: validity near 1, low JS divergence, lower instability than most baselines, and competitive diversity/interpretability. The authors also study temperature effects and handle immutable features with a masking trick. That is genuinely useful for applied explainability in credit and similar domains.

The problem is Theorem 4.1. The bound between the Gumbel-softmax density and their simplified product form is load-bearing for Equation 16, which justifies the categorical guidance update. The proof in Appendix A uses the inequality pi_i^{1-x_i} ≤ (1-x_min)^{1-x_i} (Eq. A.17). That inequality is false; a concrete counterexample is pi=(0.95,0.05), x=(0.2,0.8), x_min=0.2, where 0.95^0.8≈0.96 > 0.8^0.8≈0.84. So the upper bound derivation collapses, and with it the claim that Equation 16 is a controlled approximation. The lower bound also has some hand-wavy steps involving the Lebesgue measure of the simplex. In essence, the theoretical contribution as stated is not proven.

That doesn't sink the method entirely. The empirical story is plausibly independent of the bound—the approximation could still be good in practice, and the results suggest it is. But the paper currently overclaims. I'd recommend requesting a correction: either fix the proof, replace it with a valid bound, or drop the theorem and present Equation 16 as a heuristic with a small empirical validation of the approximation quality. Also missing are the most relevant baselines—diffusion-based counterfactual generators like TabDDPM or DiDDM-style methods—and no code is released, which makes reproducibility harder.

Who is this for? Practitioners building recourse systems on tabular data and researchers working on guided discrete diffusion. The idea has legs, but the current manuscript needs revision. A serious referee should be engaged; a desk reject would be a mistake. I'd send it to review and ask the authors to fix the proof or reframe the contribution.

Worth a read, but with a red flag on the theorem.

Referee Report

3 major / 6 minor

Summary. The paper proposes TDCE, a tabular counterfactual-explanation method that extends classifier guidance to categorical features by relaxing one-hot vectors with a Gumbel-softmax reparameterization. The reverse process is guided by classifier gradients; a product-form approximation to the Gumbel-softmax density is introduced, and Theorem 4.1 claims a KL-divergence bound between the two. Experiments on four tabular datasets compare TDCE with Wachter, CCHVAE, REVISE, CLUE, FACE, and CounterNet on validity, interpretability, diversity, stability, JS divergence, L2 distance, and runtime. The paper reports that TDCE achieves competitive or superior performance on most metrics, especially validity and interpretability, while also being faster than search-based baselines.

Significance. If the method and its supporting analysis are correct, the paper makes a useful contribution: it provides a simple, model-agnostic way to perform classifier guidance on categorical tabular features within a diffusion framework, a setting that has received less attention than image counterfactuals. The empirical tables are presented with standard deviations and the runtime comparison is informative. However, the theoretical bound is not currently reliable, the algorithm as written does not transparently implement the derived update, and the empirical comparison lacks diffusion-based baselines and uses per-dataset tuning of the temperature. These issues affect the strength of the main claims, but the underlying idea is plausible and the experimental results are promising. The paper does not provide code or a reproducibility artifact, so independent verification is not yet possible.

major comments (3)

[Appendix A, Eq. (A.17)] The proof of Theorem 4.1 relies on the inequality π_i^{1-x_i} ≤ (1-x_min)^{1-x_i}, but this is false in general. For example, with π=(0.95,0.05), x=(0.2,0.8), x_min=0.2, the left-hand side is 0.95^0.8 ≈ 0.96 while the right-hand side is 0.8^0.8 ≈ 0.84. Consequently, the upper-bound derivation leading to Eq. (A.22) collapses. Since Theorem 4.1 is the only theoretical justification for the approximation in Eq. (16), and Eq. (16) is used in deriving the guided update in Eq. (18), the theoretical guarantee for the categorical reverse process is not established. The authors should either repair the proof, provide a valid bound, or explicitly reframe Eq. (18) as a heuristic approximation supported only by experiments.
[Algorithm 1, line 7] The update `ex_cat ← µ_cat + ||µ_cat|| g_cat` does not match the derived guided density in Eq. (18), which suggests updating the logits as `log π_θ + λ g_cat`. The algorithm also omits λ. It is unclear what µ_cat represents (predicted x_0? predicted logits? the mean of a Gumbel-softmax transition?) and how it is computed from the denoiser output. If the implemented update differs from Eq. (18), the derivation and the experiments may not correspond to the same method. Please specify the exact relationship and provide the sampling step or the closed-form update that follows from Eq. (18).
[Section 5.3 and Tables 2–3] The empirical comparison is weakened by two issues. First, the temperature τ is searched per dataset over [0.1, 5] using the benchmark metrics themselves, while the baselines are reported with CARLA defaults. This can inflate the relative performance of TDCE. Second, there are no diffusion-based counterfactual baselines, despite the existence of recent discrete-diffusion guidance methods that the paper cites (e.g., [32,33]). A fairer comparison with baselines tuned on validation splits, plus at least one diffusion-based baseline, is needed to support the claim that TDCE 'outperforms popular baseline methods.'

minor comments (6)

[Theorem 4.1 vs. Appendix A.2] The upper bound in the main text uses `log[1 - x_min]`, while Appendix A, Theorem A.2 states `log x_min` and the derivation in A.22 uses `log(1 - x_min)`. These versions should be reconciled.
[Figure 3] The figure lacks axis labels and numerical values. Since the paper claims a 'tight bound,' it would help to plot the actual KL and both bounds on the same axes, or at least report the numerical values used.
[Section 4.4 and Algorithm 1] The temperature schedule (warm start, then decrease) is described in the text but not reflected in Algorithm 1. Please state the schedule explicitly or add it to the pseudocode.
[Section 5.3] Typo: 'low IM2/IM2' should read 'low IM1/IM2.' Also, 'Efficency' in the subsection heading should be 'Efficiency.'
[Section 3.2] The function `f_dn` is used in Eq. (8) before being defined. Define it at first use.
[Table 4] Caption says 'on the same computer setting'; should be 'same hardware and software settings' or similar.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the categorical guidance update is derived from an explicit Gumbel-softmax approximation and external classifier gradients, and the results are benchmarked externally.

full rationale

The paper's derivation chain is self-contained rather than circular. The guided categorical update (Eq. 18) follows from three explicit modeling choices: the Gumbel-softmax density (Eq. 13), the product-form approximation (Eq. 16/19), and a first-order Taylor expansion of the classifier (Eq. 17). None of these is fitted to the counterfactual metrics reported in Tables 2-3; the KL bound in Theorem 4.1 is a mathematical statement about the two distributions and is not a renamed version of the experimental results. The temperature τ is tuned on validation metrics, which is standard hyperparameter selection and does not make the reported counterfactual quality a prediction forced by construction. Self-citations (refs [11], [38]) appear only in related-work surveys and are not load-bearing for the method. The empirical comparison uses external baselines (Wachter, CCHVAE, REVISE, CLUE, FACE, CounterNet). The only notable issue is in the proof of Theorem 4.1: the inequality π_i^{1-x_i} ≤ (1-x_min)^{1-x_i} (Appendix A, Eq. A.17) is not valid in general, which undermines the stated upper bound; however, this is a correctness/mathematical-error concern, not a circularity, because the method does not reduce to its inputs by definition.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The method rests on standard diffusion and Gumbel-softmax machinery. Two hyperparameters (tau, lambda) are free. The key modeling axiom is the product approximation in Eq. (19), whose theoretical support (Theorem 4.1) is invalid as written.

free parameters (2)

temperature tau = searched in [0.1, 5] per dataset
Controls Gumbel-softmax sharpness; affects approximation quality, gradient variance, and performance. Selected per dataset on the reported metrics.
guidance strength lambda = not reported
Regularization hyperparameter in Eq. (18); no value or tuning details given.

assumptions (3)

domain assumption The reverse process models the conditional distribution over discrete data using the Gumbel-softmax relaxation with temperature tau.
Standard for reparameterized discrete variables, but the approximation accuracy depends on the unproven (and likely flawed) bound in Theorem 4.1.
domain assumption First-order Taylor expansion of the classifier log-likelihood in the Gumbel-softmax space (Eq. 17) yields a valid guidance direction.
Common in classifier guidance; assumes small step sizes and smooth classifier.
ad hoc to paper The diffusion model's reverse distributions p_theta(x_t|x_{t+1}) are accurately approximated by the categorical-style product in Eq. (19).
This is the approximation the paper introduces; its justification is the flawed KL bound.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Tabular Diffusion Counterfactual Explanations." pith.science (2026). https://pith.science/paper/WTRRHFU4

@misc{pith2026250900876,
  author       = {Pith},
  title        = {Pith review of: Tabular Diffusion Counterfactual Explanations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WTRRHFU4}},
  note         = {Machine review of arXiv:2509.00876}
}

abstract

Counterfactual explanations methods provide an important tool in the field of {interpretable machine learning}. Recent advances in this direction have focused on diffusion models to explain a deep classifier. However, these techniques have predominantly focused on problems in computer vision. In this paper, we focus on tabular data typical in finance and the social sciences and propose a novel guided reverse process for categorical features based on an approximation to the Gumbel-softmax distribution. Furthermore, we study the effect of the temperature $\tau$ and derive a theoretical bound between the Gumbel-softmax distribution and our proposed approximated distribution. We perform experiments on several large-scale credit lending and other tabular datasets, assessing their performance in terms of the quantitative measures of interpretability, diversity, instability, and validity. These results indicate that our approach outperforms popular baseline methods, producing robust and realistic counterfactual explanations.

Figures

Figures reproduced from arXiv: 2509.00876 by the authors.

**Figure 2.** A simulation of diffusions for the Gumbel-softmax vector over a single categorical [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗

**Figure 3.** KL divergence between the Gumbel-softmax distribution and our approximation [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Qualitative comparisons between TDCE and others methods on LCD dataset. [PITH_FULL_IMAGE:figures/full_fig_p028_4.png]

**Figure 5.** Figure 5: An analysis on the temperature τ of LCD dataset. The best JS score is achieved when τ = 0.3 with balanced IM1 and IM2 scores. increases, the counterfactual generator tends to recover the distribution of the categorical variable in the target class. However, according t…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

44 extracted references · 35 canonical work pages

[1]

M. T. Ribeiro, S. Singh, C. Guestrin, ” why should i trust you?” ex- plaining the predictions of any classifier, in: International Conference on Knowledge Discovery and Data Mining, 2016

work page 2016
[2]

Ibrahim, M

M. Ibrahim, M. Louie, C. Modarres, J. Paisley, Global explanations of neural networks: Mapping the landscape of predictions, in: AAAI/ACM Conference on AI, Ethics, and Society, 2019

work page 2019
[3]

Shrikumar, P

A. Shrikumar, P. Greenside, A. Kundaje, Learning important features through propagating activation differences, in: International Conference on Machine Learning, 2017

work page 2017
[4]

Sundararajan, A

M. Sundararajan, A. Taly, Q. Yan, Axiomatic attribution for deep networks, in: International Conference on Machine Learning, 2017. 29

work page 2017
[5]

Lundberg, S.-I

S. Lundberg, S.-I. Lee, A unified approach to interpreting model predic- tions, in: Advances in Neural Information Processing Systems, 2017

work page 2017
[6]

L. T. Liu, S. Barocas, J. Kleinberg, K. Levy, On the actionability of outcome prediction, in: AAAI Conference on Artificial Intelligence, 2024

work page 2024
[7]

L. T. Liu, S. Dean, E. Rolf, M. Simchowitz, M. Hardt, Delayed impact of fair machine learning, in: International Conference on Machine Learning, 2018

work page 2018
[8]

Agarwal, L

R. Agarwal, L. Melnick, N. Frosst, X. Zhang, B. Lengerich, R. Caruana, G. E. Hinton, Neural additive models: Interpretable machine learning with neural nets, in: Advances in Neural Information Processing Systems, 2021

work page 2021

Show all 44 references

[9]

Radenovic, A

F. Radenovic, A. Dubey, D. Mahajan, Neural basis models for inter- pretability, in: Advances in Neural Information Processing Systems, 2022

2022
[10]

Chang, R

C.-H. Chang, R. Caruana, A. Goldenberg, Node-GAM: Neural gen- eralized additive model for interpretable deep learning, International Conference on Learning Representations (2022)

2022
[11]

Zhang, B

W. Zhang, B. Barr, J. Paisley, Gaussian process neural additive models, in: AAAI Conference on Artificial Intelligence, 2024

2024
[12]

Wachter, B

S. Wachter, B. Mittelstadt, C. Russell, Counterfactual explanations without opening the black box: Automated decisions and the GDPR, Harvard Journal of Law & Tech. 31 (2017) 841. 30

2017
[13]

Poyiadzi, K

R. Poyiadzi, K. Sokol, R. Santos-Rodriguez, T. De Bie, P. Flach, F ACE: feasible and actionable counterfactual explanations, in: AAAI/ACM Conference on AI, Ethics, and Society, 2020

2020
[14]

Van Looveren, J

A. Van Looveren, J. Klaise, Interpretable counterfactual explanations guided by prototypes, in: Joint European Conference on Machine Learning and Knowledge Discovery in Databases, 2021

2021
[15]

Dandl, C

S. Dandl, C. Molnar, M. Binder, B. Bischl, Multi-objective counterfactual explanations, in: International Conference on Parallel Problem Solving from Nature, 2020

2020
[16]

Guidotti, A

R. Guidotti, A. Monreale, S. Ruggieri, D. Pedreschi, F. Turini, F. Gi- annotti, Local rule-based explanations of black box decision systems, arXiv preprint arXiv:1805.10820 (2018)

2018 arXiv
[17]

R. K. Mothilal, A. Sharma, C. Tan, Explaining machine learning clas- sifiers through diverse counterfactual explanations, in: Conference on Fairness, Accountability, and Transparency, 2020

2020
[18]

R. R. Fern´ andez, I. M. De Diego, V. Ace˜ na, A. Fern´ andez-Isabel, J. M. Moguerza, Random forest explainability using counterfactual sets, Infor- mation Fusion 63 (2020) 196–207

2020
[19]

D. P. Kingma, Auto-encoding variational Bayes, arXiv preprint arXiv:1312.6114 (2013)

2013 arXiv
[20]

Joshi, O

S. Joshi, O. Koyejo, W. Vijitbenjaronk, B. Kim, J. Ghosh, Towards realistic individual recourse and actionable explanations in black-box decision making systems, arXiv preprint arXiv:1907.09615 (2019). 31

1907 arXiv
[21]

Antor´ an, U

J. Antor´ an, U. Bhatt, T. Adel, A. Weller, J. M. Hern´ andez-Lobato, Getting a clue: A method for explaining uncertainty estimates, arXiv preprint arXiv:2006.06848 (2020)

2006 arXiv
[22]

Pawelczyk, K

M. Pawelczyk, K. Broelemann, G. Kasneci, Learning model-agnostic counterfactual explanations for tabular data, in: The Web Conference, 2020

2020
[23]

J. Ho, A. Jain, P. Abbeel, Denoising diffusion probabilistic models, in: Advances in Neural Information Processing Systems, 2020

2020
[24]

J. Song, C. Meng, S. Ermon, Denoising diffusion implicit models, arXiv preprint arXiv:2010.02502 (2020)

2010 arXiv
[25]

Dhariwal, A

P. Dhariwal, A. Nichol, Diffusion models beat GANs on image synthesis, in: Advances in Neural Information Processing Systems, 2021

2021
[26]

Augustin, V

M. Augustin, V. Boreiko, F. Croce, M. Hein, Diffusion visual counter- factual explanations, in: Advances in Neural Information Processing Systems, 2022

2022
[27]

Hoogeboom, D

E. Hoogeboom, D. Nielsen, P. Jaini, P. Forr´ e, M. Welling, Argmax flows and multinomial diffusion: Learning categorical distributions, in: Advances in Neural Information Processing Systems, 2021

2021
[28]

H. Sun, L. Yu, B. Dai, D. Schuurmans, H. Dai, Score-based continuous- time discrete diffusion models, arXiv preprint arXiv:2211.16750 (2022)

2022 arXiv
[29]

Dieleman, L

S. Dieleman, L. Sartran, A. Roshannai, N. Savinov, Y. Ganin, P. H. 32 Richemond, A. Doucet, R. Strudel, C. Dyer, C. Durkan, et al., Continuous diffusion for categorical data, arXiv preprint arXiv:2211.15089 (2022)

2022 arXiv
[30]

Kotelnikov, D

A. Kotelnikov, D. Baranchuk, I. Rubachev, A. Babenko, Tabddpm: Mod- elling tabular data with diffusion models, in: International Conference on Machine Learning, 2023

2023
[31]

Regol, M

F. Regol, M. Coates, Diffusing Gaussian mixtures for generating cate- gorical data, in: AAAI Conference on Artificial Intelligence, 2023

2023
[32]

Gruver, S

N. Gruver, S. Stanton, N. Frey, T. G. Rudner, I. Hotzel, J. Lafrance- Vanasse, A. Rajpal, K. Cho, A. G. Wilson, Protein design with guided discrete diffusion, in: Advances in Neural Information Processing Sys- tems, 2024

2024
[33]

Schiff, S

Y. Schiff, S. S. Sahoo, H. Phung, G. Wang, S. Boshar, H. Dalla-torre, B. P. de Almeida, A. Rush, T. Pierrot, V. Kuleshov, Simple guidance mech- anisms for discrete diffusion models, arXiv preprint arXiv:2412.10193 (2024)

2024 arXiv
[34]

E. Jang, S. Gu, B. Poole, Categorical reparameterization with Gumbel- softmax, in: International Conference on Learning Representations, 2017

2017
[35]

Nazabal, P

A. Nazabal, P. M. Olmos, Z. Ghahramani, I. Valera, Handling incomplete heterogeneous data using vaes, Pattern Recognition 107 (2020) 107501

2020
[36]

H. Guo, T. H. Nguyen, A. Yadav, Counternet: End-to-end training of prediction aware counterfactual explanations, in: Proceedings of the 29th 33 ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2023, pp. 577–589

2023
[37]

H. Guo, F. Jia, J. Chen, A. Squicciarini, A. Yadav, Rocoursenet: Robust training of a prediction aware recourse model, in: Proceedings of the 32nd ACM International Conference on Information and Knowledge Management, 2023, pp. 619–628

2023
[38]

Zhang, B

W. Zhang, B. Barr, J. Paisley, An interpretable deep classifier for coun- terfactual generation, in: Proceedings of the Third ACM International Conference on AI in Finance, 2022, pp. 36–43

2022
[39]

J. Su, D. V. Vargas, K. Sakurai, One pixel attack for fooling deep neural networks, IEEE Transactions on Evolutionary Computation 23 (2019) 828–841

2019
[40]

Sohl-Dickstein, E

J. Sohl-Dickstein, E. Weiss, N. Maheswaranathan, S. Ganguli, Deep unsupervised learning using nonequilibrium thermodynamics, in: Inter- national Conference on Machine Learning, 2015

2015
[41]

A. Q. Nichol, P. Dhariwal, Improved denoising diffusion probabilistic models, in: International Conference on Machine Learning, 2021

2021
[42]

C. J. Maddison, A. Mnih, Y. W. Teh, The concrete distribution: A continuous relaxation of discrete random variables, arXiv preprint arXiv:1611.00712 (2016)

2016 arXiv
[43]

Avrahami, D

O. Avrahami, D. Lischinski, O. Fried, Blended diffusion for text-driven editing of natural images, in: IEEE Conference on Computer Vision and Pattern Recognition, 2022. 34

2022
[44]

log Γ(K)τ K−1 KY i=1 Z(π)π1−exi i ex−τ −1 i KP j πj ex−τ j # (A.15) = E

M. Pawelczyk, S. Bielawski, J. v. d. Heuvel, T. Richter, G. Kasneci, CARLA: A Python library to benchmark algorithmic recourse and coun- terfactual explanation algorithms, in: Neural Information Processing Systems Track on Datasets and Benchmarks, 2021. 35 Appendix A. Proof of...

2021

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

M. T. Ribeiro, S. Singh, C. Guestrin, ” why should i trust you?” ex- plaining the predictions of any classifier, in: International Conference on Knowledge Discovery and Data Mining, 2016

work page 2016

[2] [2]

Ibrahim, M

M. Ibrahim, M. Louie, C. Modarres, J. Paisley, Global explanations of neural networks: Mapping the landscape of predictions, in: AAAI/ACM Conference on AI, Ethics, and Society, 2019

work page 2019

[3] [3]

Shrikumar, P

A. Shrikumar, P. Greenside, A. Kundaje, Learning important features through propagating activation differences, in: International Conference on Machine Learning, 2017

work page 2017

[4] [4]

Sundararajan, A

M. Sundararajan, A. Taly, Q. Yan, Axiomatic attribution for deep networks, in: International Conference on Machine Learning, 2017. 29

work page 2017

[5] [5]

Lundberg, S.-I

S. Lundberg, S.-I. Lee, A unified approach to interpreting model predic- tions, in: Advances in Neural Information Processing Systems, 2017

work page 2017

[6] [6]

L. T. Liu, S. Barocas, J. Kleinberg, K. Levy, On the actionability of outcome prediction, in: AAAI Conference on Artificial Intelligence, 2024

work page 2024

[7] [7]

L. T. Liu, S. Dean, E. Rolf, M. Simchowitz, M. Hardt, Delayed impact of fair machine learning, in: International Conference on Machine Learning, 2018

work page 2018

[8] [8]

Agarwal, L

R. Agarwal, L. Melnick, N. Frosst, X. Zhang, B. Lengerich, R. Caruana, G. E. Hinton, Neural additive models: Interpretable machine learning with neural nets, in: Advances in Neural Information Processing Systems, 2021

work page 2021