Pith. sign in

REVIEW 5 major objections 5 minor 39 references

Contrastive Token-level Explanations for Graph-based Rumour Detection

T0 review · 5 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read This paper claims that CT-LRP, a contrastive token-level LRP framework, produces higher-fidelity, sparser explanations for GNN-based rumour detection than node-level explainability baselines, with an average fidelity gain of 25.77% over…

desk verdict Sensible token-level LRP extension for GNN rumour detection, but the BERT backward pass is unspecified, Algorithm 1 has a masking bug, and the fidelity-sparsity comparison mixes units; fixable, but needs major revision. read the letter →

arxiv 2502.04366 v1 pith:MU6PEAEG submitted 2025-02-05 cs.CL cs.AIcs.LG

Daniel Wai Kit Chin , Roy Ka-Wei Lee This is my paper · ORCID
classification cs.CLcs.AIcs.LG
keywords rumourdetectiongraphneuralnetworksexplainabilitylayerwiserelevancepropagationtoken-levelexplanationscontrastivefidelity-sparsitymisinformation
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

Rumour detection models built on graph neural networks give a prediction about an event from a tree of posts, but usual explanations say which posts mattered, not which words. This paper claims that pushing layerwise relevance propagation one step further, through the text embedding function and then through a contrastive masking step, produces token-level explanations that are both more faithful and more focused than node-level explanations. The authors test the framework on three GNN rumour detectors and three public datasets, and report that token-level explanations raise fidelity and fidelity-sparsity scores across the board, with an average 25.77% fidelity gain over token-level LRP without the contrastive step. If the claim holds, explainability for rumour detection shifts from 'this post drove the decision' to 'this word in this post drove the decision,' which is the resolution fact-checkers and platform moderators need.

What carries the argument

The load-bearing mechanism is the epsilon-stabilized LRP propagation rule, $r_j = \sum_k \frac{a_j w_{jk}}{\epsilon + \sum_{j'} a_{j'} w_{j'k}} r_k$, which conserves relevance as it flows backward through the GNN and produces a node-feature attribution map $R \in \mathbb{R}^{|V|\times|D|}$. To turn that into token attributions, the paper assumes the text encoder is a composition of per-token embedding $f_{\text{embed}}$ followed by a pooling function, and for mean pooling derives the closed-form rule $r_{t_v,d} = \frac{x_{t_v,d}}{\epsilon + x_{v,d}} r_{v,d}$; summing over embedding dimensions gives each token's relevance. The contrastive refinement computes attribution maps for all classes and masks tokens that are positively relevant to the predicted class but more influential on another class, using a logit-difference test on a graph where that token's vector has been removed. The mask product yields the final class-specific token explanation.

What would settle it

Run CT-LRP on a model whose node features come from BERT, then for each token the method highlights as class-specific, remove that token from the post text, re-run the full BERT-plus-GNN pipeline, and compare the predicted-class logit change against the token's attributed relevance rank. If the highest-attribution tokens do not cause the largest logit drops, the Equation (3) decomposition is not faithful. A sharper check is to verify relevance conservation directly: replace BERT with static word vectors plus mean pooling and confirm that the sum of token relevance equals the node-level relevance; with BERT, the contextualized word vectors mean mean-pooling equality generally fails, so the check should reveal the mismatch.

Watch

Extended reading notes

Core claim

The central claim is that the class-disambiguation step is what makes token-level explanations work for GNN rumour detectors: token-level LRP alone already improves fidelity over node-level baselines, and the contrastive filter adds a further, consistent gain. The method computes an LRP attribution map for each output class, then for each token that is positively relevant in both the predicted class and another class it removes that token's vector from the pooled node embedding, re-runs the model, and keeps the token only if removing it drops the predicted-class logit more than it drops every rival-class logit; tokens with negative relevance in the predicted class are masked out as well. The surviving tokens form the explanation. Across BiGCN, EBGCN, and ClaHi-GAT trained on Twitter, Weibo, and PHEME, the framework reports higher fidelity at fixed sparsity than node-level LRP, Grad-CAM, and contrastive excitation backprop, and reports an average 66.98% higher fidelity-sparsity score than the next best baseline.

Load-bearing premise

The token-level attribution rule assumes each word is embedded independently and the post vector is exactly the average of those word vectors, but the actual encoder, BERT, produces word vectors that change with context, so Equation (3) may not describe the model being explained.

Editorial extensions

If this is right

  • If the fidelity results hold, a rumour detector's decision can be drilled down to the exact words in the source post and replies that push it toward a label, rather than only the posts that matter.
  • Because the contrastive mask filters out tokens shared with rival classes, the remaining explanations are class-specific, which should help distinguish cues for 'false rumour' from 'unverified rumour'.
  • The fidelity-sparsity gains imply explanations stay informative even when heavily pruned, which matters for large conversation trees where full token lists are unusable.
  • The framework runs one backward pass per class, so the added interpretability comes at a modest computational cost and can be applied to existing trained GNNs.
  • The authors argue the same token-level framing can be carried over to fake news detection and other misinformation tasks.

Reading between the lines

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

  • A direct way to isolate the source of the reported gain is to apply the same contrastive logit-difference mask to node-level LRP; if the mask alone explains most of the fidelity improvement, token granularity may be less important than the class-disambiguation step.
  • Because Equation (3) is exact only for static per-token embeddings plus mean pooling, a natural test is to run CT-LRP with such an encoder and compare with the BERT-based results; a large discrepancy would indicate the reported gains depend on the pooling assumption rather than on genuinely contextualized attributions.
  • The paper's own limitation note that graph inductive biases are not explained suggests a straightforward extension: combine CT-LRP's token mask with edge-level attributions to test whether propagation structure and lexical cues interact in rumour decisions.
  • A human evaluation could extend the quantitative claim: ask annotators which words in a source post are most diagnostic of a rumour label, and check whether CT-LRP's retained tokens align better than those of node-level baselines; the paper reports no such user study.
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

5 major / 5 minor

Summary. The paper proposes CT-LRP, a post hoc explainability framework that combines Layerwise Relevance Propagation (LRP) on a GNN with backpropagation through a text embedding function, aiming to produce token-level attributions for GNN-based rumour detection. The authors introduce a contrastive mask intended to keep only tokens that are specific to the predicted class, and they evaluate the method with BiGCN, EBGCN, and ClaHi-GAT on Twitter15/16, Weibo, and PHEME. They report average fidelity improvements of 25.77% over token-level LRP and 66.98% over the next best baseline on fidelity-sparsity. The central claim is that CT-LRP consistently produces high-fidelity, token-level explanations that are more granular than existing node- or edge-level graph explainability methods.

Significance. If the method worked as described, token-level explanations for GNN-based rumour detection would be a useful contribution, since existing graph explainability tools typically stop at node or edge granularity. The paper makes a concrete, falsifiable empirical claim and tests it across three model architectures and three public datasets with cross-validation, reporting standard deviations. It also extends fidelity and sparsity metrics to the token level, which is a sensible direction. However, the current manuscript has load-bearing problems: the token-level LRP rule is not valid for the actual BERT-based text encoder used in the experiments, the contrastive masking pseudocode contradicts the prose, and the fidelity-sparsity comparison mixes token units and node units. These issues prevent the reported numbers from supporting the paper's central claim as written.

major comments (5)
  1. [Section IV-C, Eq. (3)] The mean-pooling LRP rule in Eq. (3) is not the rule implied by Eq. (1). Since x_v,d is defined as (1/|T_v|) times the sum of x_{t_v,d}, the numerator should contain the same factor 1/|T_v|; omitting it breaks relevance conservation. As written, Eq. (3) is not a valid LRP propagation rule for the pooling operation described in Eq. (2).
  2. [Section IV-C and Section V-B1] The token-level computation assumes that ftext decomposes into independent per-token embeddings followed by a pooling function, but the experiments use a pre-trained multilingual BERT as ftext. BERT produces contextualized token vectors, so each x_{t_v,d} is a complex function of the entire input sequence through self-attention, residual connections, and layer normalization. The paper gives no LRP propagation rules for these BERT layers, so the quantity LRPtext(ftext, R(c), P) in Algorithm 1 is unspecified and the computation is not reproducible. Applying Eq. (3) directly to BERT output vectors does not yield a valid input-level token attribution; the authors need to either provide complete LRP rules for the transformer stack or use a genuinely non-contextual embedding model.
  3. [Algorithm 1 and Section IV-D] The masking logic in Algorithm 1 does not implement the described exclusion of negative-attribution tokens. The branch `if z(hat_y) <= 0 then Zmask <- False` is followed by an independent if/else over z(c), and the else branch assigns True regardless, so the False assignment can be overwritten. In addition, the contrastive check is executed per class inside the loop, whereas the prose requires the condition y_hat - y'_hat > y_c - y'_c to hold for all c != hat_y before retaining a token. As written, the pseudocode can retain or discard a token based on only one other class, depending on loop order.
  4. [Section V-D and Tables II-IV] The fidelity and fidelity-sparsity comparisons are not unit-consistent. For node-level baselines, sparsity is computed as 1 - m(c)/|V| and fidelity is measured by removing attributed nodes, while for CT-LRP sparsity counts tokens and fidelity is measured by removing attributed tokens. Removing one node is a much larger perturbation than removing one token, so the reported advantages in fidelity, sparsity, and fidelity-sparsity may reflect the different removal units rather than a genuinely better explanation. A controlled comparison is needed, for example by evaluating all methods with the same removal budget measured in embedded-vector mass, or by reporting a token-level baseline under the same token-removal protocol.
  5. [Section V-C and Section V-D] The fidelity protocol is described inconsistently. The metric definition says elements with attribution score greater than 0.01 are removed, while the experimental section removes elements in decreasing order of importance until a fixed sparsity level is reached. These are different procedures and could produce different results; the paper should specify exactly which protocol produced Tables II-IV.
minor comments (5)
  1. [Figure 3] The legend contains a curve labeled 'CT-LRP w/ NSC', but the abbreviation NSC is never defined or discussed in the text.
  2. [Section V-C] The c-EB baseline is described only as 'Contrastive EB' with a reference; the paper should specify how the contrastive variant is implemented and how its probabilistic Winner-Takes-All process is applied to the GNN node features.
  3. [Section I] In the contributions list, 'Boarder Impact' should be 'Broader Impact', and the phrase 'the the Russia-Ukraine war' in the introduction contains a duplicated article.
  4. [Section IV-C, Eq. (5)] The max-pooling rule in Eq. (5) is not an epsilon-stabilized LRP rule; it is a hard argmax mask that allocates all relevance to the maximum neuron. If max pooling is used in any experiment, the paper should justify this rule separately.
  5. [Section VI] The discussion states that CT-LRP 'represents a significant advancement' and 'achieves a new standard', but these claims go beyond the evidence presented, especially given the unresolved metric and implementation issues; please temper the wording.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: CT-LRP's attributions are computed from standard LRP rules plus a model-output contrastive filter, none of which is fitted to or defined by the headline fidelity results.

full rationale

I examined the derivation chain: Eq. (1) is the standard epsilon-stabilised LRP rule from [24]; Eqs. (2)/(4) define pooling; Eqs. (3)/(5) give the token relevance assignments; Algorithm 1 adds the contrastive mask using perturbed-logit comparisons. The reported quantities (fidelity, sparsity, fidelity-sparsity) are computed after this chain, not used as fitting targets inside it. The contrastive condition y_hat - y'_hat > y_c - y'_c is a per-token selection rule based on the explained model's own outputs; this makes the explanation model-centred, but that is the standard semantics of fidelity, not a circular reduction of the method's output to the metric. No parameter is fitted to the test-set fidelity scores, and no self-citation carries a load-bearing premise: [24] and [39] are external standard references for LRP and metrics. The paper's serious technical weakness is different: Eq. (3) is not a valid LRP rule for the actual ftext (pre-trained multilingual BERT), both because the 1/|T_v| factor from Eq. (2) is dropped and because BERT token vectors are contextual, with no propagation rules supplied for attention, residual, or layer-norm operations. That is a correctness/reproducibility risk, not a circularity. The Section VI limitation (graph inductive biases are not explicitly explained) is likewise a scope statement, not a circular step. I therefore find no step that reduces to its own input by construction.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard LRP conservation (assumed, not re-derived here), a decomposition assumption about BERT that is not stated, and an evaluation assumption that mixes granularity units. No parameters are fitted to the target result, which keeps circularity low, but the missing BERT backprop details make the method underspecified.

free parameters (2)
  • epsilon (LRP stabilizer) = not reported
    Small constant in Eqs. 1 and 3 to avoid division by zero; its value is not given, and it affects the sparsity of attributions.
  • fidelity attribution threshold = 0.01
    The fidelity metric removes elements with attribution score greater than 0.01; the threshold is taken from [39], not fitted, but chosen.
assumptions (4)
  • domain assumption Epsilon-stabilized LRP conserves relevance through GCN and classifier layers
    Used without modification for BiGCN, EBGCN, and ClaHi-GAT despite differing architectures, Section IV-B.
  • domain assumption Text embedding function decomposes into per-token embedding plus pooling with no cross-token interaction
    Required for Eq. 3; not stated or justified for the multilingual BERT encoder used in Section V-B1.
  • domain assumption Removing a token vector before aggregation (G' = G - x_tv) captures its causal influence on logits
    Used in Algorithm 1 to decide class membership; ignores that BERT contextualizes tokens.
  • domain assumption Fidelity at fixed sparsity is comparable across token-level and node-level explanations
    Section V-C2 redefines sparsity for CT-LRP in token units while node baselines use node units.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Contrastive Token-level Explanations for Graph-based Rumour Detection." pith.science (2026). https://pith.science/paper/MU6PEAEG

@misc{pith2026250204366,
  author       = {Pith},
  title        = {Pith review of: Contrastive Token-level Explanations for Graph-based Rumour Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MU6PEAEG}},
  note         = {Machine review of arXiv:2502.04366}
}
read the original abstract

The widespread use of social media has accelerated the dissemination of information, but it has also facilitated the spread of harmful rumours, which can disrupt economies, influence political outcomes, and exacerbate public health crises, such as the COVID-19 pandemic. While Graph Neural Network (GNN)-based approaches have shown significant promise in automated rumour detection, they often lack transparency, making their predictions difficult to interpret. Existing graph explainability techniques fall short in addressing the unique challenges posed by the dependencies among feature dimensions in high-dimensional text embeddings used in GNN-based models. In this paper, we introduce Contrastive Token Layerwise Relevance Propagation (CT-LRP), a novel framework designed to enhance the explainability of GNN-based rumour detection. CT-LRP extends current graph explainability methods by providing token-level explanations that offer greater granularity and interpretability. We evaluate the effectiveness of CT-LRP across multiple GNN models trained on three publicly available rumour detection datasets, demonstrating that it consistently produces high-fidelity, meaningful explanations, paving the way for more robust and trustworthy rumour detection systems.

Figures

Figures reproduced from arXiv: 2502.04366 by the authors.

Figure 1
Figure 1. Overview of the proposed CT-LRP framework showing the flow of information through the GNN and text embedding function. Inputs to the forward [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Token-level Explanation produced by CT-LRP. Node-level attribution is shown with darker shades of blue indicating greater attribution from that node. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Quantitative performance comparisons on three models trained on three datasets with the curves obtained by varying sparsity levels. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 30 canonical work pages

  1. [1]

    The spread of true and false news online,

    S. V osoughi, D. Roy, and S. Aral, “The spread of true and false news online,” science, vol. 359, no. 6380, pp. 1146–1151, 2018

    2018

  2. [2]

    Fake news, disinformation and misinformation in social media: a review,

    E. A ¨ımeur, S. Amri, and G. Brassard, “Fake news, disinformation and misinformation in social media: a review,” Social Network Analysis and Mining, vol. 13, no. 1, p. 30, 2023

    work page 2023

  3. [3]

    Rumors detection, verification and controlling mechanisms in online social networks: A survey,

    M. Ahsan, M. Kumari, and T. P. Sharma, “Rumors detection, verification and controlling mechanisms in online social networks: A survey,”Online Social Networks and Media , vol. 14, p. 100050, 2019

    work page 2019

  4. [4]

    Automated fact checking: Task formulations, methods and future directions,

    J. Thorne and A. Vlachos, “Automated fact checking: Task formulations, methods and future directions,” in Proceedings of the 27th International Conference on Computational Linguistics, E. M. Bender, L. Derczynski, and P. Isabelle, Eds. Santa Fe, New Mexico, USA: Association for Computational Linguistics, Aug. 2018, pp. 3346–3359. [Online]. Available: http...

    work page 2018

  5. [5]

    Information credibility on twitter,

    C. Castillo, M. Mendoza, and B. Poblete, “Information credibility on twitter,” in Proceedings of the 20th International Conference on World Wide Web, ser. WWW ’11. New York, NY , USA: Association for Computing Machinery, 2011, p. 675–684. [Online]. Available: https://doi.org/10.1145/1963405.1963500

    arXiv 2011

  6. [6]

    Automatic detection of rumor on sina weibo,

    F. Yang, Y . Liu, X. Yu, and M. Yang, “Automatic detection of rumor on sina weibo,” in Proceedings of the ACM SIGKDD Workshop on Mining Data Semantics , ser. MDS ’12. New York, NY , USA: Association for Computing Machinery, 2012. [Online]. Available: https://doi.org/10.1145/2350190.2350203

    work page arXiv 2012

  7. [7]

    Real- time rumor debunking on twitter,

    X. Liu, A. Nourbakhsh, Q. Li, R. Fang, and S. Shah, “Real- time rumor debunking on twitter,” in Proceedings of the 24th ACM International on Conference on Information and Knowledge Management, ser. CIKM ’15. New York, NY , USA: Association for Computing Machinery, 2015, p. 1867–1870. [Online]. Available: https://doi.org/10.1145/2806416.2806651

    work page arXiv 2015

  8. [8]

    Detecting rumors from microblogs with recurrent neural networks,

    J. Ma, W. Gao, P. Mitra, S. Kwon, B. J. Jansen, K.-F. Wong, and M. Cha, “Detecting rumors from microblogs with recurrent neural networks,” in Proceedings of the 25th International Joint Conference on Artificial Intelligence (IJCAI 2016) . New York, NY , USA: AAAI Press, 2016, pp. 3818–3824

    work page 2016

Show all 39 references
  1. [9]

    Turing at SemEval-2017 task 8: Sequential approach to rumour stance classification with branch- LSTM,

    E. Kochkina, M. Liakata, and I. Augenstein, “Turing at SemEval-2017 task 8: Sequential approach to rumour stance classification with branch- LSTM,” in Proceedings of the 11th International Workshop on Semantic Evaluation (SemEval-2017) , S. Bethard, M. Carpuat, M. Apidianaki, ...

    2017

  2. [10]

    Detect rumors in microblog posts using propagation structure via kernel learning,

    J. Ma, W. Gao, and K.-F. Wong, “Detect rumors in microblog posts using propagation structure via kernel learning,” in Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (ACL 2017). Vancouver, Canada: Association for Computational Linguistic...

    2017

  3. [11]

    Rumor detection on twitter with tree-structured recursive neural networks,

    ——, “Rumor detection on twitter with tree-structured recursive neural networks,” in Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (ACL 2018) . Melbourne, Australia: Association for Computational Linguistics, 2018, pp. 1980–1989

    2018

  4. [12]

    Rumor detection on social media with bi-directional graph convolutional networks,

    T. Bian, X. Xiao, T. Xu, P. Zhao, W. Huang, Y . Rong, and J. Huang, “Rumor detection on social media with bi-directional graph convolutional networks,” Proceedings of the AAAI Conference on Artificial Intelligence, vol. 34, no. 01, pp. 549–556, Apr. 2020. [Online]. Available: ...

    2020

  5. [13]

    Towards propagation uncertainty: Edge-enhanced Bayesian graph convolutional networks for rumor detection,

    L. Wei, D. Hu, W. Zhou, Z. Yue, and S. Hu, “Towards propagation uncertainty: Edge-enhanced Bayesian graph convolutional networks for rumor detection,” in Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Co...

    2021

  6. [14]

    Rumor detection on Twitter with claim-guided hierarchical graph attention networks,

    H. Lin, J. Ma, M. Cheng, Z. Yang, L. Chen, and G. Chen, “Rumor detection on Twitter with claim-guided hierarchical graph attention networks,” in Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing , M.-F. Moens, X. Huang, L. Specia, and S. W....

    2021

  7. [15]

    Fake news detection: A survey of graph neural network methods,

    H. T. Phan, N. T. Nguyen, and D. Hwang, “Fake news detection: A survey of graph neural network methods,” Applied Soft Computing , vol. 139, p. 110235, 2023

    2023

  8. [16]

    A survey on automated fact-checking,

    Z. Guo, M. Schlichtkrull, and A. Vlachos, “A survey on automated fact-checking,” Transactions of the Association for Computational Lin- guistics, vol. 10, pp. 178–206, 2022

    2022

  9. [17]

    Graph neural networks for materials science and chemistry,

    P. Reiser, M. Neubert, A. Eberhard, L. Torresi, C. Zhou, C. Shao, H. Metni, C. van Hoesel, H. Schopmans, T. Sommer et al. , “Graph neural networks for materials science and chemistry,” Communications Materials, vol. 3, no. 1, p. 93, 2022

    2022

  10. [18]

    Graph signal processing, graph neural network and graph learning on biological data: a systematic review,

    R. Li, X. Yuan, M. Radfar, P. Marendy, W. Ni, T. J. O’Brien, and P. M. Casillas-Espinosa, “Graph signal processing, graph neural network and graph learning on biological data: a systematic review,” IEEE Reviews in Biomedical Engineering , vol. 16, pp. 109–135, 2021

    2021

  11. [19]

    Graph neural networks in node classification: survey and evaluation,

    S. Xiao, S. Wang, Y . Dai, and W. Guo, “Graph neural networks in node classification: survey and evaluation,” Machine Vision and Applications, vol. 33, no. 1, p. 4, 2022

    2022

  12. [20]

    Community detection in node-attributed social networks: a survey,

    P. Chunaev, “Community detection in node-attributed social networks: a survey,” Computer Science Review , vol. 37, p. 100286, 2020

    2020

  13. [21]

    A com- prehensive survey of scene graphs: Generation and application,

    X. Chang, P. Ren, P. Xu, Z. Li, X. Chen, and A. Hauptmann, “A com- prehensive survey of scene graphs: Generation and application,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 45, no. 1, pp. 1–26, 2021

    2021

  14. [22]

    Bridging knowledge graphs to generate scene graphs,

    A. Zareian, S. Karaman, and S.-F. Chang, “Bridging knowledge graphs to generate scene graphs,” in Computer Vision – ECCV 2020, A. Vedaldi, H. Bischof, T. Brox, and J.-M. Frahm, Eds. Cham: Springer Interna- tional Publishing, 2020, pp. 606–623

    2020

  15. [23]

    Explainability in graph neural networks: A taxonomic survey,

    H. Yuan, H. Yu, S. Gui, and S. Ji, “Explainability in graph neural networks: A taxonomic survey,” IEEE transactions on pattern analysis and machine intelligence , vol. 45, no. 5, pp. 5782–5799, 2022

    2022

  16. [24]

    On pixel-wise explanations for non-linear classifier deci- sions by layer-wise relevance propagation,

    S. Bach, A. Binder, G. Montavon, F. Klauschen, K.-R. M ¨uller, and W. Samek, “On pixel-wise explanations for non-linear classifier deci- sions by layer-wise relevance propagation,” PloS one, vol. 10, no. 7, p. e0130140, 2015

    2015

  17. [25]

    Grad-cam: Visual explanations from deep networks via gradient-based localization,

    R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-cam: Visual explanations from deep networks via gradient-based localization,” in Proceedings of the IEEE International Conference on Computer Vision (ICCV) , Oct 2017

    2017

  18. [26]

    Top- down neural attention by excitation backprop,

    J. Zhang, S. A. Bargal, Z. Lin, J. Brandt, X. Shen, and S. Sclaroff, “Top- down neural attention by excitation backprop,” International Journal of Computer Vision, vol. 126, no. 10, pp. 1084–1102, 2018

    2018

  19. [27]

    Review and comparison of methods to study the contribution of variables in artificial neural network models,

    M. Gevrey, I. Dimopoulos, and S. Lek, “Review and comparison of methods to study the contribution of variables in artificial neural network models,” Ecological modelling, vol. 160, no. 3, pp. 249–264, 2003

    2003

  20. [28]

    ”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 , ser. KDD ’16. New York, NY , USA: Association for Comput...

    2016

  21. [29]

    A unified approach to interpreting model predictions,

    S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” in Advances in Neural Information Processing Systems, I. Guyon, U. V . Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, Eds., vol. 30. Curran Associates, Inc., 2017

    2017

  22. [30]

    Gnnex- plainer: Generating explanations for graph neural networks,

    Z. Ying, D. Bourgeois, J. You, M. Zitnik, and J. Leskovec, “Gnnex- plainer: Generating explanations for graph neural networks,” Advances in neural information processing systems , vol. 32, 2019

    2019

  23. [31]

    Parameterized explainer for graph neural network,

    D. Luo, W. Cheng, D. Xu, W. Yu, B. Zong, H. Chen, and X. Zhang, “Parameterized explainer for graph neural network,” Advances in neural information processing systems , vol. 33, pp. 19 620–19 631, 2020

    2020

  24. [32]

    Zorro: Valid, sparse, and stable explanations in graph neural networks,

    T. Funke, M. Khosla, M. Rathee, and A. Anand, “Zorro: Valid, sparse, and stable explanations in graph neural networks,” IEEE Transactions on Knowledge and Data Engineering , vol. 35, no. 8, pp. 8687–8698, 2022

    2022

  25. [33]

    Graphlime: Local interpretable model explanations for graph neural networks,

    Q. Huang, M. Yamada, Y . Tian, D. Singh, and Y . Chang, “Graphlime: Local interpretable model explanations for graph neural networks,”IEEE Transactions on Knowledge and Data Engineering , vol. 35, no. 7, pp. 6968–6972, 2022

    2022

  26. [34]

    Deep inside convolutional networks: Visualising image classification models and saliency maps,

    K. Simonyan, A. Vedaldi, and A. Zisserman, “Deep inside convolutional networks: Visualising image classification models and saliency maps,” arXiv preprint arXiv:1312.6034 , 2013

    2013 arXiv

  27. [35]

    Striving for simplicity: The all convolutional net,

    J. T. Springenberg, A. Dosovitskiy, T. Brox, and M. Riedmiller, “Striving for simplicity: The all convolutional net,” ICLR 2015 Workshop, 2014

    2015

  28. [36]

    Visualizing and understanding convolu- tional networks,

    M. D. Zeiler and R. Fergus, “Visualizing and understanding convolu- tional networks,” in Computer Vision – ECCV 2014 , D. Fleet, T. Pajdla, B. Schiele, and T. Tuytelaars, Eds. Cham: Springer International Publishing, 2014, pp. 818–833

    2014

  29. [37]

    Analysing how people orient to and spread rumours in social media by looking at conversational threads,

    A. Zubiaga, G. W. S. Hoi, M. Liakata, R. Procter, and P. Tolmie, “Analysing how people orient to and spread rumours in social media by looking at conversational threads,” PLoS ONE, vol. 11, 2015. [Online]. Available: https://api.semanticscholar.org/CorpusID:14134758

    2015

  30. [38]

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

    J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre- training of deep bidirectional transformers for language understanding,” in Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technolog...

    2019

  31. [39]

    Explainability methods for graph convolutional neural networks,

    P. E. Pope, S. Kolouri, M. Rostami, C. E. Martin, and H. Hoffmann, “Explainability methods for graph convolutional neural networks,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2019. Daniel Wai Kit Chin completed his Bachel...

    2019

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.