Pith. sign in

REVIEW 4 major objections 5 minor 60 references

Enhancing Treatment Effect Estimation via Active Learning: A Counterfactual Covering Perspective

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read The error of active treatment-effect estimation is governed by factual and counterfactual covering radii, so label acquisition is a covering problem.

desk verdict Genuinely new covering perspective with solid empirical gains, but the theory as written doesn't hold together: the core bound assumes zero training loss, and the approximation proofs have gaps. read the letter →

arxiv 2505.05242 v1 pith:4HVVNCQY submitted 2025-05-08 cs.LG

Hechuan Wen , Tong Chen , Mingming Gong , Li Kheng Chai , Shazia Sadiq , Hongzhi Yin This is my paper · ORCID
classification cs.LG
keywords activelearningtreatmenteffectestimationcounterfactualcoveringradiusfactualPEHEcoveragemaximizationcore-setcausalinference
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

When treatment outcomes are expensive to label, the paper claims, the error of a treatment-effect estimator trained on a small actively chosen subset is governed, with high probability, by how well the labeled points cover both treatment groups in covariate space. The reducible part of the risk bound is a sum of two kinds of covering radii: a factual radius within each treatment group and a counterfactual radius from each group to the other group's points. This reframes label acquisition as a covering problem: spend the labeling budget on points that shrink the largest radius. The paper proves this bound, gives a greedy radius-reduction algorithm with a factor-two approximation under idealized conditions, and proposes FCCM, a coverage-maximization variant for partially overlapping distributions. The stakes are practical: the same inequality says what information a labeling budget should buy to improve individualized treatment-effect estimates.

What carries the argument

The load-bearing objects are the factual covering radius $\delta(t,t)$ and the counterfactual covering radius $\delta(t,1-t)$: the radius of the smallest ball centered at the labeled samples from treatment group $t$ whose union covers all pool samples of group $t$ (factual) or of the opposite group $1-t$ (counterfactual). These radii enter linearly in Theorem 3.4, so shrinking them is the only model-independent lever on the risk bound; the $\kappa_H$ discrepancy term is fixed once the pool is given. The second mechanism is the greedy reduction procedure, which at each step selects the point that reduces the largest of the four radii, reducing to the k-center problem for the factual radii and to a cross-group covering problem for the counterfactual radii. FCCM replaces radius minimization with coverage maximization on a weighted directed graph whose edges connect each candidate center to points inside its factual and counterfactual balls; it repeatedly picks the highest-degree vertex and removes covered edges, which is the max-coverage machinery that yields the $1-1/e$ guarantee.

What would settle it

On the TOY setting, run FCCM with a deliberately non-interpolating estimator (for example, a regularized neural network with non-negligible weight decay or early stopping). Record the sum of the four covering radii, the training loss on the labeled set, and the PEHE at each acquisition step; if radii decrease substantially while training loss stays well above zero and PEHE does not follow the bound's reductions, the claim that covering radius controls risk would be refuted in a concrete case.

Watch

Extended reading notes

Core claim

The paper's central claim is Theorem 3.4: for an i.i.d. pool of $n$ covariate samples, with probability at least $1-\gamma$, the subset generalization gap — the difference between the population PEHE risk and the empirical training loss on the selected labeled subset — is at most $\sum_{t\in\{0,1\}} \kappa_t(\delta(t,t)+\delta(t,1-t)) + 2\kappa_H + \sqrt{L_l^2\log(1/\gamma)/(2n)}$, where $\kappa_H$ is a distributional discrepancy (an integral probability metric) between the treated and control groups. Since the four covering radii decrease as the labeled set grows, the gap is $O(\delta(1,1)) + O(\delta(1,0)) + O(\delta(0,0)) + O(\delta(0,1))$. The paper therefore claims that the model-independent, reducible part of treatment-effect estimation risk under active learning is exactly a covering problem: label points so every pool point is close to a labeled point of its own group and close to a labeled point of the opposite group. It further proves that a greedy k-center-style radius-reduction algorithm is within a factor of two of the optimal sum of radii under strong ignorability, and that FCCM, which maximizes coverage under fixed radii, approximates full coverage to within $1-1/e$ under its assumptions.

Load-bearing premise

The bound assumes the trained estimator achieves zero training loss on the selected labeled subset; if the model does not interpolate the labeled outcomes, the covering-radius terms no longer control the subset generalization gap, so the acquisition objective is not proven to reduce risk.

Editorial extensions

If this is right

  • If Theorem 3.4 is correct, any acquisition strategy that only maximizes diversity inside one treatment group leaves $\delta(t,1-t)$ uncontrolled, so the risk bound need not shrink even as the labeled set grows.
  • Corollary 3.5 implies that, with a fixed pool, the subset generalization gap is $O(\delta(1,1)) + O(\delta(1,0)) + O(\delta(0,0)) + O(\delta(0,1))$; the largest radius is the right next acquisition target.
  • Theorem 4.1 says that under strong ignorability the greedy radius-reduction algorithm returns a sum of radii within a factor of two of optimal, so an idealized balanced distribution admits a near-optimal k-center-style acquisition rule.
  • Theorem 4.4 says FCCM is a $(1-1/e)$-approximation to full factual and counterfactual coverage under its assumptions, providing a tractable surrogate for radius reduction on realistic partially overlapping distributions.
  • In the paper's experiments, FCCM reaches lower PEHE than both general active learning baselines and treatment-effect-specific baselines on TOY, IBM, and CMNIST, and its acquisitions concentrate on high-density overlap regions.

Reading between the lines

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

  • One could apply the same radius objective to more than two treatments or to continuous treatments by defining $\delta(t,t')$ for every pair of treatment regions; the paper only develops the binary case.
  • A testable extension is to track the four covering radii on the pool during acquisition and check whether they track PEHE; the paper does not implement such a diagnostic.
  • Because the bound assumes interpolating models, a natural follow-up is an acquisition rule that adds a weighted empirical training-loss term for non-interpolating estimators; the paper does not derive one.
  • For non-overlapping treatment groups, FCCM's acquisition is challenged; a latent-space variant that aligns the groups before covering would sacrifice some model independence, and the paper flags this as an open limitation.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper studies active learning for treatment effect estimation under a labeling budget, formalizing the subset selection problem as minimization of a PEHE risk upper bound. Its central theoretical contribution is Theorem 3.4, which bounds the subset generalization gap by the sum of four data-dependent quantities: factual covering radii δ(t,t) and counterfactual covering radii δ(t,1−t), plus a distributional discrepancy term and a finite-sample term. Based on this bound, the authors propose Algorithm 1, a greedy radius-reduction method with a 2-OPT guarantee, and Algorithm 2 (FCCM), a coverage-maximization method claimed to give a (1−1/e)-approximation. They evaluate FCCM on TOY, IBM, and CMNIST datasets against general and treatment-effect-specific active learning baselines, reporting lower PEHE in most settings and including ablation and sensitivity studies.

Significance. If the theoretical claims were sound, the paper would contribute a useful model-independent principle for data-efficient treatment effect estimation: acquisition should reduce factual and counterfactual covering radii rather than merely reduce uncertainty or diversity. The empirical evaluation is reasonably thorough, with multiple datasets, standard baselines, ablations, and released code, and the visualization of acquisition behavior supports the qualitative explanation of FCCM's gains. However, the main theoretical bridge from covering radii to PEHE risk, Theorem 3.4, relies on a zero-training-loss assumption that the experimental estimators do not satisfy, and the approximation guarantees in Theorems 4.1 and 4.4 contain gaps. These issues are central to the paper's stated contributions and need to be repaired before the theoretical framing can be accepted.

major comments (4)
  1. The proof of Theorem 3.4 drops the empirical training loss through the zero-training-loss assumption introduced before Eq. (5) and used again in Eq. (13g). In particular, the subset generalization gap is equated to (1/n)Σ_i ξ(x_i; f_S) only when (1/|S|)Σ_{j∈S} l(x_j,y_j,t_j; f_S) = 0, and the transition to Eq. (13c) also drops the expected subset losses ε_S^t. If the trained estimator does not interpolate the labeled outcomes, both terms are positive and the displayed expression is no longer an upper bound on the actual subset generalization gap. The estimators used in the experiments, DUE-DNN and DUE-CNN, are deep kernel models with inducing points, dropout, and spectral norm regularization, for which exact interpolation of the selected training points is not expected. Since Theorem 3.4 is the step that converts the covering-radius objective into a guarantee on treatment effect estimation risk, the paper needs either an argument that these estimators interpolate the selected data, an explicit training-loss term carried through the bound, or an alternative derivation that does not rely on zero training loss.
  2. The proof of Lemma A.9 assumes that the proxy collection S-tilde_{1−t} has an identical counterpart in D_t, stating that strong ignorability guarantees an 'identical set' S_{(t,1−t)} ∈ D_t matching the proxy collection. Strong ignorability gives overlap of the covariate supports of the two treatment groups; it does not imply that every proxy point has an exact copy in the opposite treatment group. If the nearest factual point satisfies d(a^t, a^{1−t}) > 0, the equality d(u^{1−t}, S_{(t,1−t)}) = d(u^{1−t}, S-tilde_{1−t}) in Eq. (43a) fails, and the subsequent chain does not yield the claimed 2-OPT bound for the counterfactual radius. The theorem should state the exact-match condition explicitly or incorporate the approximation error of the nearest factual point into the bound.
  3. The equality P(A) = (1/2)P(A_F^{t=1} ∪ A_CF^{t=1}) + (1/2)P(A_F^{t=0} ∪ A_CF^{t=0}) does not follow from the definition of P(A) in Eq. (9). For each group t, the identity P(A_F^t) + P(A_CF^t) = 2 P(A_F^t ∪ A_CF^t) requires that the factual and counterfactual cover sets are disjoint and that |D_t| = |D_{1−t}|, conditions that are neither stated nor generally true for the datasets used. Without this equality, the proof does not establish that maximizing the four individual coverage fractions in Eq. (9) is equivalent to maximizing the two union-coverage terms that the greedy analysis actually addresses.
  4. The approximation guarantee in Theorem 4.4 is stated for an equally weighted graph with unscaled out-degree, but Algorithm 2 as defined uses counterfactual edge weight α and a scaled out-degree coefficient c(x) = ζ(x)(1−ζ(x)). The proof does not track either α or the scaling coefficient, so the (1−1/e) guarantee does not apply to the algorithm as implemented and evaluated. The paper should either prove the guarantee for the weighted, scaled objective actually optimized by FCCM or clearly state that the theorem covers a simplified variant and justify empirically that the modification preserves the approximation behavior.
minor comments (5)
  1. The symbol D is used both for the full pool set in Eq. (1) and for the training dataset in the definition of the estimator f_D at the start of Section 2; this conflation should be resolved with separate notation for the labeled and unlabeled pools.
  2. The Hoeffding inequality is written as an equality; it should be an upper bound, P(·) ≤ exp(−2nε²/L_l²), and the variable ε here is overloaded with the risk notation used elsewhere.
  3. The assumption states that the squared loss l is bounded by L_l, but a squared loss on unbounded outcomes is not bounded; the boundedness condition should be stated as an explicit restriction on the support of the outcomes or on the hypothesis class.
  4. The comment in the sketch version that 'a is not labeled' is confusing because the algorithm outputs S as the set to be labeled; the distinction between the proxy point a′ and the factual query a should be stated more clearly in both the sketch and the full pseudocode.
  5. Several figure captions are inconsistent, including two figures numbered 'Figure 4' and a caption for the CMNIST sensitivity plot that says 'TOY'; these should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Theorem 3.4 is derived from explicit Lipschitz and zero-training-loss assumptions rather than fitted to the empirical PEHE.

full rationale

The derivation of Theorem 3.4 is self-contained. Equation (4) decomposes the PEHE risk into a generalization error, a subset generalization gap, and an empirical training loss; the zero-training-loss assumption is stated explicitly before Eq. (5) and is used again in Eq. (13g) only to drop the empirical loss term. The remaining bound is obtained through proved lemmas (A.2, A.3, A.5) that translate Lipschitz continuity and the integral-probability-metric constant into the factual and counterfactual covering radii, rather than through any fitted parameter or by assuming the desired conclusion. The approximation guarantees for Algorithms 1 and 2 (Theorems 4.1 and 4.4) reduce to standard k-center and max-coverage arguments, not to the empirical treatment-effect results. The only self-citation, Wen et al. (2025) for MACAL, appears as related work and as a source of estimator hyperparameters; it does not carry the proof of any theorem or the claim that covering radii control PEHE. The zero-training-loss premise may be empirically unrealistic for the DUE estimators used in the experiments, but that is a validity concern about an explicitly stated assumption, not a circular step: the paper does not relabel a fitted quantity as a prediction, and the validation tuning of δ and α is ordinary model selection rather than fitting the theory to the outcome. No step in the derivation chain reduces by construction to its own inputs.

Assumptions & free parameters 2 free parameters · 6 assumptions · 0 invented entities

The theoretical results rest on standard causal assumptions plus Lipschitz-type conditions and the zero-training-loss idealization. FCCM's approximation guarantee additionally assumes a full-cover solution exists. The empirical behavior depends on validation-tuned hyperparameters δ and α.

free parameters (2)
  • covering radius δ = 0.11 (TOY, IBM), 0.40-0.50 (CMNIST)
    FCCM's acquisition graph uses a fixed covering radius for factual and counterfactual balls; the radius is tuned per dataset on the validation set around a 95% coverage threshold (Section 4.3, Table 4).
  • counterfactual edge weight α = 2.5
    Weight for counterfactual edges in the acquisition graph; chosen by validation search over {1.0, 2.5, 5.0} (Appendix C.6, Table 4).
assumptions (6)
  • domain assumption Consistency: only one potential outcome is observed per unit given treatment t (Assumption 2.2).
    Standard causal inference assumption used to connect observed outcomes to potential outcomes.
  • domain assumption Strong Ignorability: {Y_t=0, Y_t=1} ⊥ t | x and 0 < p(t=1|x) < 1 (Assumption 2.3).
    Required for identifiability of ITE and used in the proof of Theorem 4.1 for the counterfactual covering radius reduction.
  • domain assumption Lipschitz continuity of conditional densities p_t(y|x) and squared loss, with bounded loss L_l (Assumption 3.2).
    Needed to relate loss at an unlabeled point to loss at its nearest labeled center, yielding the covering radius terms in Theorem 3.4.
  • domain assumption Existence of constant κ such that h_f(x,t) = (1/κ) l(f(x,t)) ∈ H (Assumption 3.3).
    Borrowed from Shalit et al. 2017; used to bound the counterfactual loss via an IPM term κ_H.
  • domain assumption Zero training loss on the selected subset S (Section 3, Eq. (5); Appendix A.1 Eq. (13g)).
    The proof of Theorem 3.4 equates the subset generalization gap with the empirical counterfactual loss by dropping the training loss term; this is an interpolation assumption that rarely holds exactly for noisy outcomes.
  • domain assumption There exists an optimal solution S* that achieves full factual and counterfactual coverage with the fixed radii (Assumption 4.3).
    Needed for the (1−1/e) approximation guarantee of FCCM; fails when treatment groups have minimal overlap.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Enhancing Treatment Effect Estimation via Active Learning: A Counterfactual Covering Perspective." pith.science (2026). https://pith.science/paper/4HVVNCQY

@misc{pith2026250505242,
  author       = {Pith},
  title        = {Pith review of: Enhancing Treatment Effect Estimation via Active Learning: A Counterfactual Covering Perspective},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4HVVNCQY}},
  note         = {Machine review of arXiv:2505.05242}
}
read the original abstract

Although numerous complex algorithms for treatment effect estimation have been developed in recent years, their effectiveness remains limited when handling insufficiently labeled training sets due to the high cost of labeling the effect after treatment, e.g., expensive tumor imaging or biopsy procedures needed to evaluate treatment effects. Therefore, it becomes essential to actively incorporate more high-quality labeled data, all while adhering to a constrained labeling budget. To enable data-efficient treatment effect estimation, we formalize the problem through rigorous theoretical analysis within the active learning context, where the derived key measures -- \textit{factual} and \textit{counterfactual covering radius} determine the risk upper bound. To reduce the bound, we propose a greedy radius reduction algorithm, which excels under an idealized, balanced data distribution. To generalize to more realistic data distributions, we further propose FCCM, which transforms the optimization objective into the \textit{Factual} and \textit{Counterfactual Coverage Maximization} to ensure effective radius reduction during data acquisition. Furthermore, benchmarking FCCM against other baselines demonstrates its superiority across both fully synthetic and semi-synthetic datasets.

Figures

Figures reproduced from arXiv: 2505.05242 by the authors.

Figure 1
Figure 1. Visualization of the factual covering (FC) and the counterfactual covering (CFC) on the dataset by the acquired samples from each group. Note that each covering is constrained by the full coverage on the desired dataset with the minimum radius. generalization gap ∆ is upper-bounded as [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Visuals of the radius reduction and the descent of the Bound under ideal and realistic data distributions by Algorithm 1. counterfactual covering radii decline synchronously (five plots fully overlap with each other) under a quick risk con￾vergence to zero, because Eq. (8) is reduced to the simple k-Center problem (where Assumption 2.3 surely satisfies) which guarantees a 2-OP T approximation and the greedy nature o… view at source ↗
Figure 3
Figure 3. Visualization of the high coverage by Algorithm 2 on CMNIST, and reduction gain over mean coverage loss by Algo￾rithm 2 when compared to Algorithm 1. To solve (11), we propose a greedy solution – factual and counterfactual coverage maximization (FCCM) in Algo￾rithm 2. Specifically, FCCM constructs a weighted graph G with the node V by the entire covariate matrix X ∈ D, and each node v t ∈ Dt builds the directed edge… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Estimating a smaller range for the covering radius δ around the 95% coverage threshold by Algorithm 2. formance compared to Algorithm 1 is accessible in [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: All plots are the mean values averaged from 10 simulations associated with the standard deviation as the error bar. Note that all models at 0% exhibit the same performance given the fixed estimators and are thus neglected. The performance under 2% granularity is presen…
Figure 6
Figure 6. Figure 6: Visualizations of the toy dataset distribution, and the actual acquisition of the data by FCCM, µρBALD, and BAIT. The size of the data point and color from (b) to (c) is adjusted to its associated density, with deeper color representing higher density and vice versa. d…
Figure 7
Figure 7. Figure 7: Visualization of the factual covering (dashed manifold) and the counterfactual covering (solid manifold) on the local neighbor￾hood from the center x. Note that the zero-neighbor scenario is omitted as it is the least preferred to maximize the coverage, Thus, the scali…
Figure 8
Figure 8. Figure 8: High dimensional data projected into 2-dimensional space via t-SNE, with colors indicating overlapping, group t = 1, and group t = 0. Left: entire data distribution; Mid: distribution on group t = 1; Right: distribution on group t = 0. C.5. Practicability of the Assump…
Figure 3
Figure 3. Figure 3: 2% Increment on CMNIST Dataset. (c) 2% increment on CMNIST dataset C.8. Sensitivity Study Note that the acquisition on treatment sample t = 1 is insensitive on δ(0,0) and δ(0,1) in our setting, as all control samples (t = 0) are seen. For α, our setting of α = 2.5 has …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

60 extracted references · 45 canonical work pages

  1. [1]

    Sample constrained treatment effect estimation

    Addanki, R., Arbour, D., Mai, T., Musco, C., and Rao, A. Sample constrained treatment effect estimation. NeurIPS, 35: 0 5417--5430, 2022

    work page 2022

  2. [2]

    Alaa, A. M. and Van Der Schaar, M. Bayesian inference of individualized treatment effects using multi-task gaussian processes. NeurIPS, 2017

    work page 2017

  3. [3]

    Gone fishing: Neural active learning with fisher embeddings

    Ash, J., Goel, S., Krishnamurthy, A., and Kakade, S. Gone fishing: Neural active learning with fisher embeddings. NeurIPS, 34: 0 8927--8939, 2021

    work page 2021

  4. [4]

    T., Zhang, C., Krishnamurthy, A., Langford, J., and Agarwal, A

    Ash, J. T., Zhang, C., Krishnamurthy, A., Langford, J., and Agarwal, A. Deep batch active learning by diverse, uncertain gradient lower bounds. In International Conference on Learning Representations, 2019

    work page 2019

  5. [5]

    Doubly Robust Causal Effect Estimation under Networked Interference via Targeted Learning

    Chen, W., Cai, R., Yang, Z., Qiao, J., Yan, Y., Li, Z., and Hao, Z. Doubly robust causal effect estimation under networked interference via targeted learning. arXiv preprint arXiv:2405.03342, 2024

    work page Pith review arXiv 2024

  6. [6]

    Task-specific experimental design for treatment effect estimation

    Connolly, B., Moore, K., Schwedes, T., Adam, A., Willis, G., Feige, I., and Frye, C. Task-specific experimental design for treatment effect estimation. In International Conference on Machine Learning, pp.\ 6384--6401. PMLR, 2023

    work page 2023

  7. [7]

    J., Cunningham, W

    Cook, W. J., Cunningham, W. H., Pulleyblank, W. R., and Schrijver, A. Combinatorial Optimization. Springer, 1998

    work page 1998

  8. [8]

    Active learning for personalizing treatment

    Deng, K., Pineau, J., and Murphy, S. Active learning for personalizing treatment. In 2011 IEEE Symposium on Adaptive Dynamic Programming and Reinforcement Learning (ADPRL), pp.\ 32--39. IEEE, 2011

    work page 2011

Show all 60 references
  1. [9]

    Lecture 4: Approximation algorithms - vertex cover and set cover

    Dinitz, M. Lecture 4: Approximation algorithms - vertex cover and set cover. https://www.cs.jhu.edu/ mdinitz/classes/ApproxAlgorithms/Spring2019/Lectures/lecture4.pdf, 2019

    2019

  2. [10]

    and Kashima, H

    Fujii, K. and Kashima, H. Budgeted stream-based active learning via adaptive submodular maximization. NeurIPS, 29, 2016

    2016

  3. [11]

    Deep bayesian active learning with image data

    Gal, Y., Islam, R., and Ghahramani, Z. Deep bayesian active learning with image data. In International conference on machine learning, pp.\ 1183--1192. PMLR, 2017

    2017

  4. [12]

    Ghadiri, M., Arbour, D., Mai, T., Musco, C., and Rao, A. B. Finite population regression adjustment and non-asymptotic guarantees for treatment effect estimation. NeurIPS, 36, 2024

    2024

  5. [13]

    Hill, J. L. Bayesian nonparametric modeling for causal inference. Journal of Computational and Graphical Statistics, 20 0 (1): 0 217--240, 2011

    2011

  6. [14]

    u ller, D., Zaverkin, V., K \

    Holzm \"u ller, D., Zaverkin, V., K \"a stner, J., and Steinwart, I. A framework and benchmark for deep batch active learning for regression. Journal of Machine Learning Research, 24 0 (164): 0 1--81, 2023

    2023

  7. [15]

    Imbens, G. W. and Rubin, D. B. Causal inference in statistics, social, and biomedical sciences. Cambridge University Press, 2015

    2015

  8. [16]

    Identifying causal-effect inference failure with uncertainty-aware models

    Jesson, A., Mindermann, S., Shalit, U., and Gal, Y. Identifying causal-effect inference failure with uncertainty-aware models. NeurIPS, 33: 0 11637--11649, 2020

    2020

  9. [17]

    Quantifying ignorance in individual-level causal-effect estimates under hidden confounding

    Jesson, A., Mindermann, S., Gal, Y., and Shalit, U. Quantifying ignorance in individual-level causal-effect estimates under hidden confounding. In International Conference on Machine Learning, pp.\ 4829--4838. PMLR, 2021 a

    2021

  10. [18]

    Causal-bald: Deep bayesian active learning of outcomes to infer treatment-effects from observational data

    Jesson, A., Tigas, P., van Amersfoort, J., Kirsch, A., Shalit, U., and Gal, Y. Causal-bald: Deep bayesian active learning of outcomes to infer treatment-effects from observational data. NeurIPS, 34: 0 30465--30478, 2021 b

    2021

  11. [19]

    Deepmatch: Balancing deep covariate representations for causal inference using adversarial training

    Kallus, N. Deepmatch: Balancing deep covariate representations for causal inference using adversarial training. In International Conference on Machine Learning, pp.\ 5067--5077. PMLR, 2020

    2020

  12. [20]

    Stochastic batch acquisition: A simple baseline for deep active learning

    Kirsch, A., Farquhar, S., Atighehchian, P., Jesson, A., Branchaud-Charron, F., and Gal, Y. Stochastic batch acquisition: A simple baseline for deep active learning. arXiv preprint arXiv:2106.12059, 2021

    2021 arXiv

  13. [21]

    and Longbotham, R

    Kohavi, R. and Longbotham, R. Online controlled experiments and a/b tests. Encyclopedia of machine learning and data mining, pp.\ 1--11, 2015

    2015

  14. [22]

    The mnist database of handwritten digits

    LeCun, Y. The mnist database of handwritten digits. http://yann. lecun. com/exdb/mnist/, 1998

    1998

  15. [23]

    Estimating treatment effects under heterogeneous interference

    Lin, X., Zhang, G., Lu, X., Bao, H., Takeuchi, K., and Kashima, H. Estimating treatment effects under heterogeneous interference. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases, pp.\ 576--592. Springer, 2023

    2023

  16. [24]

    Treatment effect estimation under unknown interference

    Lin, X., Zhang, G., Lu, X., and Kashima, H. Treatment effect estimation under unknown interference. In Pacific-Asia Conference on Knowledge Discovery and Data Mining, pp.\ 28--42. Springer, 2024

    2024

  17. [25]

    Scalable individual treatment effect estimator for large graphs

    Lin, X., Bao, H., Cui, Y., Takeuchi, K., and Kashima, H. Scalable individual treatment effect estimator for large graphs. Machine Learning, 114 0 (1): 0 1--19, 2025

    2025

  18. [26]

    M., Sontag, D., Zemel, R., and Welling, M

    Louizos, C., Shalit, U., Mooij, J. M., Sontag, D., Zemel, R., and Welling, M. Causal effect inference with deep latent-variable models. NeurIPS, 30, 2017

    2017

  19. [27]

    Learning causal effects on hypergraphs

    Ma, J., Wan, M., Yang, L., Li, J., Hecht, B., and Teevan, J. Learning causal effects on hypergraphs. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pp.\ 1202--1212, 2022

    2022

  20. [28]

    and Tresp, V

    Ma, Y. and Tresp, V. Causal inference under networked interference and intervention policy enhancement. In International Conference on Artificial Intelligence and Statistics, pp.\ 3700--3708. PMLR, 2021

    2021

  21. [29]

    Government policy experiments and the ethics of randomization

    MacKay, D. Government policy experiments and the ethics of randomization. Philosophy & Public Affairs, 48 0 (4): 0 319--352, 2020

    2020

  22. [30]

    Causality

    Pearl, J. Causality. Cambridge university press, 2009

    2009

  23. [31]

    u ger, K., Hillebrecht, A., Pons-K \

    Pilat, C., Frech, T., Wagner, A., Kr \"u ger, K., Hillebrecht, A., Pons-K \"u hnemann, J., Scheibelhut, C., B \"o deker, R.-H., and Mooren, F.-C. Exploring effects of a natural combination medicine on exercise-induced inflammatory immune response: A double-blind rct. Scandinav...

    2015

  24. [32]

    Pinsler, R., Gordon, J., Nalisnick, E., and Hern \'a ndez-Lobato, J. M. Bayesian batch active learning as sparse subset approximation. NeurIPS, 32, 2019

    2019

  25. [33]

    Budgeted heterogeneous treatment effect estimation

    Qin, T., Wang, T.-Z., and Zhou, Z.-H. Budgeted heterogeneous treatment effect estimation. In International Conference on Machine Learning, pp.\ 8693--8702. PMLR, 2021

    2021

  26. [34]

    Linked causal variational autoencoder for inferring paired spillover effects

    Rakesh, V., Guo, R., Moraffah, R., Agarwal, N., and Liu, H. Linked causal variational autoencoder for inferring paired spillover effects. In Proceedings of the 27th ACM International Conference on Information and Knowledge Management, pp.\ 1679--1682, 2018

    2018

  27. [35]

    B., Chen, X., and Wang, X

    Ren, P., Xiao, Y., Chang, X., Huang, P.-Y., Li, Z., Gupta, B. B., Chen, X., and Wang, X. A survey of deep active learning. ACM computing surveys (CSUR), 54 0 (9): 0 1--40, 2021

    2021

  28. [36]

    M., Rotnitzky, A., and Zhao, L

    Robins, J. M., Rotnitzky, A., and Zhao, L. P. Estimation of regression coefficients when some regressors are not always observed. Journal of the American statistical Association, 89 0 (427): 0 846--866, 1994

    1994

  29. [37]

    Rosenbaum, P. R. and Rubin, D. B. The central role of the propensity score in observational studies for causal effects. Biometrika, 70 0 (1): 0 41--55, 1983

    1983

  30. [38]

    Scott, D. W. Multivariate density estimation: theory, practice, and visualization. John Wiley & Sons, 2015

    2015

  31. [39]

    and Savarese, S

    Sener, O. and Savarese, S. Active learning for convolutional neural networks: A core-set approach. In International Conference on Learning Representations, 2018

    2018

  32. [40]

    Active learning literature survey

    Settles, B. Active learning literature survey. Computer Sciences Technical Report, 2009

    2009

  33. [41]

    D., and Sontag, D

    Shalit, U., Johansson, F. D., and Sontag, D. Estimating individual treatment effect: generalization bounds and algorithms. In International Conference on Machine Learning, pp.\ 3076--3085. PMLR, 2017

    2017

  34. [42]

    Adapting neural networks for the estimation of treatment effects

    Shi, C., Blei, D., and Veitch, V. Adapting neural networks for the estimation of treatment effects. NeurIPS, 32, 2019

    2019

  35. [43]

    Benchmarking framework for performance-evaluation of causal inference analysis

    Shimoni, Y., Yanover, C., Karavani, E., and Goldschmnidt, Y. Benchmarking framework for performance-evaluation of causal inference analysis. arXiv preprint arXiv:1802.05046, 2018

    2018 arXiv

  36. [44]

    Smith, K. On the standard deviations of adjusted and interpolated values of an observed polynomial function and its constants and the guidance they give towards a proper choice of the distribution of observations. Biometrika, 12 0 (1/2): 0 1--85, 1918

    1918

  37. [45]

    Active learning for decision-making from imbalanced observational data

    Sundin, I., Schulam, P., Siivola, E., Vehtari, A., Saria, S., and Kaski, S. Active learning for decision-making from imbalanced observational data. In International conference on machine learning, pp.\ 6046--6055. PMLR, 2019

    2019

  38. [46]

    W., Kwok, J

    Tsang, I. W., Kwok, J. T., Cheung, P.-M., and Cristianini, N. Core vector machines: Fast svm training on very large data sets. Journal of Machine Learning Research, 6 0 (4), 2005

    2005

  39. [47]

    On feature collapse and deep kernel learning for single forward pass uncertainty

    Van Amersfoort, J., Smith, L., Jesson, A., Key, O., and Gal, Y. On feature collapse and deep kernel learning for single forward pass uncertainty. arXiv preprint arXiv:2102.11409, 2021

    2021 arXiv

  40. [48]

    Vapnik, V. N. An overview of statistical learning theory. IEEE transactions on neural networks, 10 0 (5): 0 988--999, 1999

    1999

  41. [49]

    Optimal transport for treatment effect estimation

    Wang, H., Fan, J., Chen, Z., Li, H., Liu, W., Liu, T., Dai, Q., Wang, Y., Dong, Z., and Tang, R. Optimal transport for treatment effect estimation. NeurIPS, 36, 2024

    2024

  42. [50]

    Active learning via query synthesis and nearest neighbour search

    Wang, L., Hu, X., Yuan, B., and Lu, J. Active learning via query synthesis and nearest neighbour search. Neurocomputing, 147: 0 426--434, 2015

    2015

  43. [51]

    K., Sadiq, S., and Yin, H

    Wen, H., Chen, T., Ye, G., Chai, L. K., Sadiq, S., and Yin, H. Progressive generalization risk reduction for data-efficient causal effect estimation. In Proceedings of 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining, volume V.1, pp.\ 1575--1586, 2025

    2025

  44. [52]

    G., Hu, Z., Salakhutdinov, R., and Xing, E

    Wilson, A. G., Hu, Z., Salakhutdinov, R., and Xing, E. P. Deep kernel learning. In Artificial intelligence and statistics, pp.\ 370--378. PMLR, 2016

    2016

  45. [53]

    Wolf, G. W. Facility location: concepts, models, algorithms and case studies, 2011

    2011

  46. [54]

    Pool-based sequential active learning for regression

    Wu, D. Pool-based sequential active learning for regression. IEEE transactions on neural networks and learning systems, 30 0 (5): 0 1348--1359, 2018

    2018

  47. [55]

    Representation learning for treatment effect estimation from observational data

    Yao, L., Li, S., Li, Y., Huai, M., Gao, J., and Zhang, A. Representation learning for treatment effect estimation from observational data. NeurIPS, 31, 2018

    2018

  48. [56]

    Active learning through a covering lens

    Yehuda, O., Dekel, A., Hacohen, G., and Weinshall, D. Active learning through a covering lens. NeurIPS, 35: 0 22354--22367, 2022

    2022

  49. [57]

    Ganite: Estimation of individualized treatment effects using generative adversarial nets

    Yoon, J., Jordon, J., and Van Der Schaar, M. Ganite: Estimation of individualized treatment effects using generative adversarial nets. In International Conference on Learning Representations, 2018

    2018

  50. [58]

    Zhan, X., Wang, Q., Huang, K.-h., Xiong, H., Dou, D., and Chan, A. B. A comparative survey of deep active learning. arXiv preprint arXiv:2203.13450, 2022

    2022 arXiv

  51. [59]

    Learning overlapping representations for the estimation of individualized treatment effects

    Zhang, Y., Bellot, A., and Schaar, M. Learning overlapping representations for the estimation of individualized treatment effects. In International Conference on Artificial Intelligence and Statistics, pp.\ 1005--1014. PMLR, 2020

    2020

  52. [60]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.