Pith. sign in

REVIEW 4 major objections 8 minor 43 references

Transparent Neighborhood Approximation for Text Classifier Explanation

T0 review · 4 major / 8 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 XPROB constructs text-explanation neighborhoods by recursive, transparent probability-based editing rather than black-box generators, and reports competitive quality with superior stability.

desk verdict A transparent, fast n-gram editing alternative to generator-based text explainers, with solid competitive results, an overclaimed stability edge, and a valid but non-fatal concern about the edit objective's grammar claims. read the letter →

arxiv 2411.16251 v1 pith:M6Y2UTV5 submitted 2024-11-25 cs.CL cs.LG

Yi Cai , Arthur Zimek , Eirini Ntoutsi , Gerhard Wunder This is my paper · ORCID
classification cs.CLcs.LG
keywords explainableAItextclassificationexplanationneighborhoodconstructionmodel-agnosticexplanationsprobability-basededitingn-gramcontextsgenerator-basedexplainersfeatureattribution
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

The paper claims that transparent, deterministic text editing can replace black-box neural generators in building the synthetic neighborhoods that model-agnostic text explainers rely on. Its proposed method, XPROB, inserts or replaces words from the text being explained into counterfactual prototypes at positions chosen by local n-gram probabilities estimated from a plain-text corpus, then recursively applies the same edit rule. The authors argue that this substitution is not a sacrifice: XPROB matches the explanation quality of the best generator-based competitor on correctness, completeness, and compactness across two datasets and two classifier architectures, while being dozens to hundreds of times faster and more stable across similar inputs. Because every generated neighbor is the result of an explicit, auditable edit, the explanation pipeline itself stays transparent.

What carries the argument

The central object is the probability-based edit operation $O(w, \hat{x})$ of Eq. 2, which selects an insertion or replacement position by maximizing the product of a left-context conditional probability $P_{\text{pre}}$ and a right-context conditional probability $P_{\text{suc}}$, estimated by n-gram counts in the retained corpus $X_p$, divided by an edit-length penalty $e^{j-i}$ to favor small changes. It is what carries the argument because it makes every generated neighbor a deterministic, auditable function of explicit corpus statistics, eliminating the black-box generator while keeping neighbors close to the natural text manifold. Around it sits the recursive editing scheme of Algorithm 1, which iterates edited texts as new prototypes until the neighborhood is filled, and the tf-idf prototype selection that enforces locality.

What would settle it

Run XPROB with a deliberately small or domain-mismatched corpus $X_p$ (say, 2,000 texts of unrelated genre) on explicands full of rare words, and compare surrogate fidelity against XPROAX on the same inputs; if the fidelity gap widens in proportion to the explicand's missing n-gram coverage, the claim that probability-based editing preserves neighborhood quality at small corpus sizes would fail.

Watch

Extended reading notes

Core claim

The paper's central claim is that substituting the generator-based construction process with recursive probability-based editing yields an explainer, XPROB, whose explanation quality is competitive with neural-generator explainers and whose neighborhood construction is fully transparent and more controllable. The decisive mechanism is the edit objective: for each word $w$ of the explicand and each prototype counterfactual $\hat{x}$, XPROB chooses the operation that maximizes $P_{\text{pre}}(w|\hat{x}^i_{i-n+1})\,P_{\text{suc}}(w|\hat{x}^{j+n-1}_{j})\,/\,e^{j-i}$, where the probabilities are corpus-count estimates of the word's compatibility with the local left and right n-gram contexts, and the denominator penalizes long deletions. Recursively applying this rule transitions from counterfactual to explicand and yields realistic neighbors without training or running a generator. The paper reports that on Yelp and Amazon review data, for both BERT and LSTM classifiers, XPROB matches XPROAX, the strongest generator-based competitor, on completeness and compactness, and that its deterministic construction produces lower attribution variance across similar contexts, which the authors credit to its controllability.

Load-bearing premise

The load-bearing premise is that n-gram counts from a 20,000-text corpus faithfully represent natural text contexts, so the edit objective's product of left and right conditional probabilities really picks the most natural place for a word; where those contexts are missing, edits become infeasible or unnatural and the neighborhood no longer reflects the classifier's local behavior.

Editorial extensions

If this is right

  • Explanation neighborhoods can be constructed without training or running a generator, cutting per-entry time cost to about one second versus five to several hundred seconds for the generator-based competitors.
  • Every synthetic neighbor is traceable to an explicit edit operation and to corpus counts, making the construction process auditable in a way that latent-space sampling is not.
  • XPROB's deterministic edits yield more stable attributions across similar contexts: on the stability test, its averaged attribution deviations for adjectives and nouns are the lowest among the compared explainers.
  • Completeness and compactness match XPROAX, the best generator-based competitor, and clearly exceed LIME, showing that transparent editing does not sacrifice explanation quality.
  • XPROB's explanation quality depends on its external corpus in a direct and predictable way, and it remains competitive with a 5,000-text corpus even when the generator baseline needs 200,000 texts.
  • The recurring editing rule transfers to any discrete sequence with local contexts, so the same machinery could build neighborhoods for code, log lines, or multi-class text settings, with counterfactuals redefined as low-confidence instances.

Reading between the lines

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

  • A testable extension the paper does not pursue: interpolated n-gram smoothing in place of raw corpus counts could reduce the corpus size XPROB needs, since Eq. 2 currently gives any unseen context the same minimal epsilon.
  • The transparency claim implies a direct human audit: users could be shown pairs of XPROB edits and generator outputs and asked which operation is easier to verify, a comparison the paper does not run.
  • Because the edit rule operates on any discrete sequence with local contexts, XPROB's scheme should transfer to code, log lines, or tabular text fields; the paper only demonstrates it on short review texts.
  • The paper's binary-class assumption is not a hard limit: redefining counterfactuals as low-confidence instances, as the conclusion notes, would let the same machinery serve multi-class settings, but that adaptation is left untested.
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

4 major / 8 minor

Summary. The paper proposes XPROB, a model-agnostic explanation method for text classifiers that constructs local neighborhoods by recursive probability-based editing. Instead of using a neural generator to synthesize neighboring texts, XPROB integrates tokens from the explicand into real counterfactual prototypes under a local n-gram objective (Eq. 2), with prototypes selected by tf-idf closeness from an external corpus Xp. The method is evaluated on Yelp and Amazon review datasets with BERT and LSTM classifiers, comparing against LIME, XSPELLS, ABELE, and XPROAX. Reported results show competitive fidelity, completeness, and compactness, lower time cost, and a stability experiment suggesting low attribution variance. The paper argues that the deterministic, rule-based editing process is more transparent and controllable than generator-based neighborhood construction.

Significance. If the empirical claims hold, the contribution is valuable: XPROB offers a simple, deterministic, and considerably faster alternative to generator-based neighborhood construction while producing explanations of comparable quality to the stronger baseline XPROAX. The method is reproducible in principle, and the authors provide source code. The evaluation is broad in scope—two real-world datasets, two classifiers, four baselines, and several complementary metrics—and the dependency analysis (Section 4.5) is a thoughtful addition. The main advertised benefits are transparency and stability, and these are only partially supported by the evidence. The paper does not contain machine-checked proofs or parameter-free derivations, but the algorithmic construction is explicit and auditable. The central 'competitive performance' claim is defensible from the reported tables, whereas the stronger 'superior stability' and 'grammatical correctness' claims need revision.

major comments (4)
  1. [Section 3.1, Eq. (2)] The objective in Eq. (2) is not a valid estimate of the sequence probability defined in Eq. (1). With n=1, which is used in all experiments (Section 4.1), Ppre(w|left) is a forward bigram conditional and Psuc(w|right) is a backward bigram conditional; their product does not equal the joint probability of the manipulated sequence, even under a Markov assumption. Consequently, the claim that 'Implementing manipulation as outlined in Eq. 1 ensures grammatical correctness' is unsupported. The qualitative examples in Table 2 actually show ungrammatical outputs, e.g., 'excellent a beginning french students' and 'the were very excellent desserts', which suggest the factorization can select off-manifold edits. The paper should either temper the grammaticality claim, justify the product as a heuristic with empirical validation, or add a human/automatic grammaticality evaluation of the generated neighborhoods.
  2. [Section 4.4, Table 6] The 'superior stability' claim is not fully supported by the reported data. In Table 6, ABELE has a lower averaged adjective deviation (0.062) than XPROB (0.082), and the paper's explanation that ABELE's low deviation comes from the smaller scale of its attribution scores is post hoc; no scale-normalized comparison is provided. XPROB does show the lowest noun deviation (0.019) and competitive adjective deviation, but no statistical significance tests accompany the comparisons. The conclusion should be revised to a more modest statement about low attribution deviation, and the analysis should include a scale-adjusted stability measure or a significance test.
  3. [Section 4.5, Tables 7–9] The dependency comparison between XPROAX and XPROB is not controlled. XPROAX is trained on a 200k-entry corpus while XPROB uses 5k–80k entries, and the analysis is limited to one dataset/model pair (Amazon LSTM) and two metrics. The statement that XPROB exhibits 'direct but less sensitive dependency' on the volume of the corpus is not directly supported, because no comparison shows how XPROAX's performance degrades with corpus size under the same evaluation. To substantiate the 'more controllable' and 'less sensitive' claims, the authors should either measure XPROAX with varying generator training sizes or explicitly reframe the results as demonstrating resource efficiency rather than differential sensitivity.
  4. [Section 4.1, hyperparameters] The paper fixes several hyperparameters—n=1, epsilon, the edit-length penalty base e, neighborhood population p=400, prototype count k=80, locality kernel width sigma, and diversity weight lambda—and notes that hyperparameter impact studies are omitted due to space limitations. Since the method's selling point is a transparent and controllable construction process, the absence of any sensitivity analysis for these choices leaves the robustness of the reported results unexamined. In particular, n controls the context size and is set to the smallest possible value; a brief study varying n would strengthen the practical claims.
minor comments (8)
  1. [Abstract] The acronym XPROB is used in the abstract but defined only in the introduction; consider defining it at first mention.
  2. [Table 2] In the second input row, the classifier is labeled 'b(·)' but the text elsewhere uses f(·); this is likely a typo and should be corrected.
  3. [Section 3.1] The phrase 'Albeit not explicitly given' is awkward; suggest rewording to 'Although not written explicitly' or similar.
  4. [Footnote 1] The source code URL contains a duplicated scheme: 'https://https://github.com/caiy0220/XPROB' should be 'https://github.com/caiy0220/XPROB'.
  5. [References] Reference [43] appears corrupted: the proceedings name includes 'frye2020shapley' and the entry should be repaired.
  6. [Algorithm 1] The set update on line 8, 'S = Snew − Snew ∩ N (x)', is not standard notation; please use set difference and intersection symbols explicitly, e.g., S ← Snew \ (Snew ∩ N(x)).
  7. [Section 4.4] The claim that XPROB's near-zero noun attributions 'accurately reflect their non-informative nature' relies on the selection criterion (confidence near 0.5 for single-word inputs) as a proxy; this connection should be acknowledged more explicitly.
  8. [Section 4.1] The generator training hyperparameters (epochs, learning rate, etc.) are not reported; the paper states they follow the original paper, but including them would improve reproducibility for the baseline comparisons.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: XPROB's derivation is self-contained; the only self-citation is a baseline (XPROAX) that is not load-bearing, and Eq. 2's factorization is an approximation concern rather than a circular reduction.

full rationale

The derivation chain is self-contained. Given an explicand x and a retained corpus Xp, XPROB defines edits by maximizing a sequence probability (Eq. 1), approximates that objective with a product of local n-gram conditional probabilities estimated by corpus counts (Eq. 2), selects prototypes as tf-idf-near counterfactuals from Xp, recursively edits them to build a neighborhood, then trains a linear surrogate with cosine-distance weights and reads attributions from the surrogate coefficients. None of these quantities is fitted to the evaluation outcome or to the claimed conclusion: the 'transparent' property is a design characterization of using explicit count-based edits, not a derived prediction. The 'competitive performance' claim is checked against external baselines on held-out test sets using fidelity, R2, completeness, compactness, and a template-based stability test, so the central claim is empirically falsifiable rather than true by construction. The only self-citation is to XPROAX [9], which is used as a competitor and as the source of the DAAE configuration for the generator-based baselines; the construction of XPROB itself does not depend on that citation. The independence-style factorization in Eq. 2 is a modeling approximation that could be questioned on validity grounds, but it is not circular: no equation in the paper reduces to its own input or to the target result by construction. The paper even acknowledges the greedy nature of recursive editing in Section 5, which is a limitation, not a circular step. Overall, the paper's load-bearing claims have independent empirical content, so circularity is minimal.

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

The method introduces no new physical or formal entities. Its central claims rest on modeling choices: n-gram context size, smoothing, edit-length penalty, kernel width, diversity weight, and prototype count, plus assumptions about corpus representativeness and the factorization of edit probabilities.

free parameters (6)
  • n-gram context size n = 1 for both datasets
    Chosen by hand because review texts are short; no sensitivity analysis over n is reported.
  • epsilon and edit validity threshold = epsilon = 1/(|Xp|+1); threshold = epsilon^2
    Smoothing constant assigned to absent n-gram contexts; value is derived from corpus size, not from a principled estimator or comparison.
  • edit-length penalty base = e in the factor e^{-(j-i)}
    Soft constraint that discourages long edits; the specific base is chosen without ablation.
  • neighborhood population and prototype count = p = 400, k = 80
    Set across all test cases; the hyperparameter impact study is omitted due to space limitations.
  • locality kernel width sigma = not reported
    Gaussian kernel width in Section 3.3 is needed to weight neighborhood instances; no value or tuning is given.
  • diversity weight lambda = not reported
    Balances closeness and diversity in instance-level explanation selection; no value or tuning is given.
assumptions (6)
  • domain assumption The product Ppre(w|preceding) * Psuc(w|succeeding) is a valid proxy for the probability of editing a word into context.
    Eq. 2 multiplies independently estimated left and right conditional probabilities; no single generative model guarantees this factorization.
  • domain assumption Local n-gram context, with n=1 here, is sufficient to preserve grammaticality and semantic plausibility of edited texts.
    The paper states this balance in Section 3.1 but does not validate it across domains or text lengths.
  • domain assumption tf-idf cosine distance selects appropriate counterfactual prototypes.
    Section 3.2 uses tf-idf to measure proximity, which reflects lexical overlap but may not capture semantic counterfactual quality.
  • domain assumption The retained corpus Xp is sampled from the target text domain and contains sufficient counterfactuals for each explicand.
    Sections 3.1 and 3.2 rely on Xp both for probability estimates and for prototype candidates; a skewed or small corpus degrades the method.
  • domain assumption A linear regressor trained on the weighted neighborhood faithfully reveals the black box's local behavior.
    Section 3.3 adopts the standard LIME-style surrogate assumption without additional justification for text data.
  • domain assumption Recursive editing from real counterfactuals toward the explicand produces a gradual transition that highlights the decision boundary.
    The paper argues this in Section 3.2, but the greedy per-word optimization may not globally trace a clean boundary.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Transparent Neighborhood Approximation for Text Classifier Explanation." pith.science (2026). https://pith.science/paper/M6Y2UTV5

@misc{pith2026241116251,
  author       = {Pith},
  title        = {Pith review of: Transparent Neighborhood Approximation for Text Classifier Explanation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/M6Y2UTV5}},
  note         = {Machine review of arXiv:2411.16251}
}
read the original abstract

Recent literature highlights the critical role of neighborhood construction in deriving model-agnostic explanations, with a growing trend toward deploying generative models to improve synthetic instance quality, especially for explaining text classifiers. These approaches overcome the challenges in neighborhood construction posed by the unstructured nature of texts, thereby improving the quality of explanations. However, the deployed generators are usually implemented via neural networks and lack inherent explainability, sparking arguments over the transparency of the explanation process itself. To address this limitation while preserving neighborhood quality, this paper introduces a probability-based editing method as an alternative to black-box text generators. This approach generates neighboring texts by implementing manipulations based on in-text contexts. Substituting the generator-based construction process with recursive probability-based editing, the resultant explanation method, XPROB (explainer with probability-based editing), exhibits competitive performance according to the evaluation conducted on two real-world datasets. Additionally, XPROB's fully transparent and more controllable construction process leads to superior stability compared to the generator-based explainers.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 41 canonical work pages

  1. [1]

    A survey on deep learning: Algorithms, techniques, and applications,

    S. Pouyanfar, S. Sadiq, Y . Yan, H. Tian, Y . Tao, M. P. Reyes, M.-L. Shyu, S.-C. Chen, and S. S. Iyengar, “A survey on deep learning: Algorithms, techniques, and applications,” ACM Computing Surveys (CSUR), vol. 51, no. 5, pp. 1–36, 2018

    work page 2018

  2. [2]

    Freely scalable and reconfigurable optical hardware for deep learning,

    L. Bernstein, A. Sludds, R. Hamerly, V . Sze, J. Emer, and D. Englund, “Freely scalable and reconfigurable optical hardware for deep learning,” Scientific Reports, vol. 11, no. 1, pp. 1–12, 2021

    work page 2021

  3. [3]

    European union regulations on algorithmic decision-making and a “right to ex- planation

    B. Goodman and S. Flaxman, “European union regulations on algorithmic decision-making and a “right to ex- planation”,” AI Magazine, vol. 38, no. 3, pp. 50–57, 2017

    work page 2017

  4. [4]

    Machine learning interpretability: A survey on methods and metrics,

    D. V . Carvalho, E. M. Pereira, and J. S. Cardoso, “Machine learning interpretability: A survey on methods and metrics,” Electronics, vol. 8, no. 8, p. 832, 2019

    work page 2019

  5. [5]

    “Why should I trust you?

    M. T. Ribeiro, S. Singh, and C. Guestrin, ““Why should I trust you?” explaining the predictions of any classifier,” in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016, pp. 1135–1144. 12 Transparent Neighborhood Approximation for Text Classifier Explanation A PREPRINT

    work page 2016

  6. [6]

    A unified approach to interpreting model predictions,

    S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” Advances in Neural Infor- mation Processing Systems, vol. 30, pp. 4765–4774, 2017

    work page 2017

  7. [7]

    Defining locality for surrogates in post-hoc interpretablity,

    T. Laugel, X. Renard, M.-J. Lesot, C. Marsala, and M. Detyniecki, “Defining locality for surrogates in post-hoc interpretablity,” in Workshop on Human Interpretability in Machine Learning (WHI)-International Conference on Machine Learning (ICML), 2018

    work page 2018

  8. [8]

    RISE: Randomized input sampling for explanation of black-box models,

    V . Petsiuk, A. Das, and K. Saenko, “RISE: Randomized input sampling for explanation of black-box models,” in British Machine Vision Conference, 2018

    work page 2018

Show all 43 references
  1. [9]

    XPROAX-local explanations for text classification with progressive neigh- borhood approximation,

    Y . Cai, A. Zimek, and E. Ntoutsi, “XPROAX-local explanations for text classification with progressive neigh- borhood approximation,” in 2021 IEEE 8th International Conference on Data Science and Advanced Analytics (DSAA). IEEE, 2021, pp. 1–10

    2021

  2. [10]

    Explaining short text classification with diverse synthetic exemplars and counter-exemplars,

    O. Lampridis, L. State, R. Guidotti, and S. Ruggieri, “Explaining short text classification with diverse synthetic exemplars and counter-exemplars,”Machine Learning, vol. 112, no. 11, pp. 4289–4322, 2023

    2023

  3. [11]

    Explaining instance classifications with interactions of subsets of feature values,

    E. ˇStrumbelj, I. Kononenko, and M. R. ˇSikonja, “Explaining instance classifications with interactions of subsets of feature values,” Data & Knowledge Engineering, vol. 68, no. 10, pp. 886–904, 2009

    2009

  4. [12]

    Algorithms to estimate shapley value feature attributions,

    H. Chen, I. C. Covert, S. M. Lundberg, and S.-I. Lee, “Algorithms to estimate shapley value feature attributions,” Nature Machine Intelligence, vol. 5, no. 6, pp. 590–601, 2023

    2023

  5. [13]

    Towards hierarchical importance attribution: Explaining composi- tional semantics for neural sequence models,

    X. Jin, Z. Wei, J. Du, X. Xue, and X. Ren, “Towards hierarchical importance attribution: Explaining composi- tional semantics for neural sequence models,” in International Conference on Learning Representations, 2019

    2019

  6. [14]

    True to the model or true to the data?

    H. Chen, J. D. Janizek, S. Lundberg, and S.-I. Lee, “True to the model or true to the data?” in Workshop on Human Interpretability in Machine Learning (WHI)-International Conference on Machine Learning (ICML) , 2020

    2020

  7. [15]

    On locality of local explanation models,

    S. Ghalebikesabi, L. Ter-Minassian, K. DiazOrdaz, and C. C. Holmes, “On locality of local explanation models,” Advances in Neural Information Processing Systems, vol. 34, pp. 18 395–18 407, 2021

    2021

  8. [16]

    Shapley explainability on the data manifold,

    C. Frye, D. de Mijolla, T. Begley, L. Cowton, M. Stanley, and I. Feige, “Shapley explainability on the data manifold,” in International Conference on Learning Representations, 2020

    2020

  9. [17]

    Explaining sentiment classification with synthetic exemplars and counter-exemplars,

    O. Lampridis, R. Guidotti, and S. Ruggieri, “Explaining sentiment classification with synthetic exemplars and counter-exemplars,” in International Conference on Discovery Science. Springer, 2020, pp. 357–373

    2020

  10. [18]

    Generating sentences from a con- tinuous space,

    S. Bowman, L. Vilnis, O. Vinyals, A. Dai, R. Jozefowicz, and S. Bengio, “Generating sentences from a con- tinuous space,” in Proceedings of The 20th SIGNLL Conference on Computational Natural Language Learning, 2016, pp. 10–21

    2016

  11. [19]

    Black box explanation by learning image exemplars in the latent feature space,

    R. Guidotti, A. Monreale, S. Matwin, and D. Pedreschi, “Black box explanation by learning image exemplars in the latent feature space,” in Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Springer, 2019, pp. 189–205

    2019

  12. [20]

    ALIME: Autoencoder based approach for local interpretability,

    S. M. Shankaranarayana and D. Runje, “ALIME: Autoencoder based approach for local interpretability,” in International Conference on Intelligent Data Engineering and Automated Learning. Springer, 2019, pp. 454– 463

    2019

  13. [21]

    CausaLM: Causal model explanation through counterfactual language models,

    A. Feder, N. Oved, U. Shalit, and R. Reichart, “CausaLM: Causal model explanation through counterfactual language models,” Computational Linguistics, vol. 47, no. 2, pp. 333–386, 2021

    2021

  14. [22]

    Beyond trivial coun- terfactual explanations with diverse valuable explanations,

    P. Rodriguez, M. Caccia, A. Lacoste, L. Zamparo, I. Laradji, L. Charlin, and D. Vazquez, “Beyond trivial coun- terfactual explanations with diverse valuable explanations,” inProceedings of the IEEE/CVF International Con- ference on Computer Vision, 2021, pp. 1056–1065

    2021

  15. [23]

    Explainable artificial intelligence (XAI) 2.0: A manifesto of open challenges and inter- disciplinary research directions,

    L. Longo, M. Brcic, F. Cabitza, J. Choi, R. Confalonieri, J. Del Ser, R. Guidotti, Y . Hayashi, F. Herrera, A. Holzinger et al., “Explainable artificial intelligence (XAI) 2.0: A manifesto of open challenges and inter- disciplinary research directions,” Information Fusion, p. ...

    2024

  16. [24]

    Language models are few-shot learners,

    T. Brown, B. Mann, N. Ryder, M. Subbiah, J. D. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell et al., “Language models are few-shot learners,” Advances in Neural Information Processing Systems, vol. 33, pp. 1877–1901, 2020

    1901

  17. [25]

    A causal framework for explaining the predictions of black-box sequence-to- sequence models,

    D. Alvarez-Melis and T. Jaakkola, “A causal framework for explaining the predictions of black-box sequence-to- sequence models,” in Proceedings of the 2017 Conference on Empirical Methods in Natural Language Process- ing, 2017, pp. 412–421

    2017

  18. [26]

    Jurafsky, Speech & language processing

    D. Jurafsky, Speech & language processing. Pearson Education India, 2000

    2000

  19. [27]

    C. D. Manning, Introduction to information retrieval. Syngress Publishing, 2008. 13 Transparent Neighborhood Approximation for Text Classifier Explanation A PREPRINT

    2008

  20. [28]

    Sentence-BERT: Sentence embeddings using siamese BERT-networks,

    N. Reimers and I. Gurevych, “Sentence-BERT: Sentence embeddings using siamese BERT-networks,” in Pro- ceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th Interna- tional Joint Conference on Natural Language Processing (EMNLP-IJCNLP),...

    2019

  21. [29]

    Salient region detection by integrating intrinsic and extrinsic cues without prior information,

    J. Ma, J. Li, Z. Li, and J. Jiao, “Salient region detection by integrating intrinsic and extrinsic cues without prior information,” Journal of Engineering Science & Technology Review, vol. 10, no. 3, 2017

    2017

  22. [30]

    Image saliency: From intrinsic to extrinsic context,

    M. Wang, J. Konrad, P. Ishwar, K. Jing, and H. Rowley, “Image saliency: From intrinsic to extrinsic context,” in Proceedings of the 2011 IEEE Conference on Computer Vision and Pattern Recognition, 2011, pp. 417–424

    2011

  23. [31]

    Diversity in machine learning,

    Z. Gong, P. Zhong, and W. Hu, “Diversity in machine learning,” IEEE Access, vol. 7, pp. 64 323–64 350, 2019

    2019

  24. [32]

    Style transfer from non-parallel text by cross-alignment,

    T. Shen, T. Lei, R. Barzilay, and T. Jaakkola, “Style transfer from non-parallel text by cross-alignment,” in Proceedings of the 31st International Conference on Neural Information Processing Systems , 2017, pp. 6833– 6844

    2017

  25. [33]

    Hidden factors and hidden topics: Understanding rating dimensions with review text,

    J. McAuley and J. Leskovec, “Hidden factors and hidden topics: Understanding rating dimensions with review text,” in Proceedings of the 7th ACM Conference on Recommender systems, 2013, pp. 165–172

    2013

  26. [34]

    BERT: Pre-training of deep bidirectional transformers for language understanding,

    J. D. M.-W. C. Kenton and L. K. Toutanova, “BERT: Pre-training of deep bidirectional transformers for language understanding,” in Proceedings of NAACL-HLT, 2019, pp. 4171–4186

    2019

  27. [35]

    Long short-term memory,

    S. Hochreiter and J. Schmidhuber, “Long short-term memory,”Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997

    1997

  28. [36]

    Educating text autoencoders: Latent representation guidance via denoising,

    T. Shen, J. Mueller, R. Barzilay, and T. Jaakkola, “Educating text autoencoders: Latent representation guidance via denoising,” in International Conference on Machine Learning. PMLR, 2020, pp. 8719–8729

    2020

  29. [37]

    Towards complementary explanations using deep neu- ral networks,

    W. Silva, K. Fernandes, M. J. Cardoso, and J. S. Cardoso, “Towards complementary explanations using deep neu- ral networks,” in Understanding and Interpreting Machine Learning in Medical Image Computing Applications. Springer, 2018, pp. 133–140

    2018

  30. [38]

    From anecdotal evidence to quantitative evaluation methods: A systematic review on evaluating explainable ai,

    M. Nauta, J. Trienes, S. Pathak, E. Nguyen, M. Peters, Y . Schmitt, J. Schl¨otterer, M. van Keulen, and C. Seifert, “From anecdotal evidence to quantitative evaluation methods: A systematic review on evaluating explainable ai,” ACM Computing Surveys, vol. 55, no. 13s, pp. 1–42, 2023

    2023

  31. [39]

    A survey of methods for explaining black box models,

    R. Guidotti, A. Monreale, S. Ruggieri, F. Turini, F. Giannotti, and D. Pedreschi, “A survey of methods for explaining black box models,” ACM Computing Surveys, vol. 51, no. 5, pp. 1–42, 2018

    2018

  32. [40]

    Comparing automatic and human evaluation of local explanations for text classification,

    D. Nguyen, “Comparing automatic and human evaluation of local explanations for text classification,” in 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Association for Computational Linguistics...

    2018

  33. [41]

    Evaluating the visualization of what a deep neural network has learned,

    W. Samek, A. Binder, G. Montavon, S. Lapuschkin, and K.-R. M ¨uller, “Evaluating the visualization of what a deep neural network has learned,”IEEE Transactions on Neural Networks and Learning Systems, vol. 28, no. 11, pp. 2660–2673, 2016

    2016

  34. [42]

    Gender bias in coreference resolution,

    R. Rudinger, J. Naradowsky, B. Leonard, and B. Van Durme, “Gender bias in coreference resolution,” in Pro- ceedings of NAACL-HLT, 2018, pp. 8–14

    2018

  35. [43]

    Adversarially regularized autoencoders,

    J. Zhao, Y . Kim, K. Zhang, A. Rush, and Y . LeCun, “Adversarially regularized autoencoders,” in International Conference on Machine frye2020shapley. PMLR, 2018, pp. 5902–5911. 14

    2018

Pith tools

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