Pith. sign in

REVIEW 3 major objections 3 minor 41 references

Improving Interpretability and Accuracy in Neuro-Symbolic Rule Extraction Using Class-Specific Sparse Filters

T0 review · 3 major / 3 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A class-specific sparsity loss that trains CNN filters to fire near 0 or 1 reduces neuro-symbolic rule-set size by 53% and lifts accuracy by 9%, closing the gap to the original CNN to about 3%.

desk verdict Novel sparsity loss with class-specific binarization targets, but the headline 9%/53% gains lack variance reporting and a direct test of the claimed root cause. read the letter →

arxiv 2501.16677 v1 pith:HSGYGTYP submitted 2025-01-28 cs.CV cs.AI

Parth Padalkar , Jaeseong Lee , Shiyi Wei , Gopal Gupta This is my paper · ORCID
classification cs.CVcs.AI
keywords neuro-symbolicruleextractionCNNinterpretabilitysparsitylossclass-specificfiltersbinarizationlogicprogrammingimageclassification
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 the accuracy loss in neuro-symbolic image classification—where a CNN's last-layer filter outputs are converted to 0/1 and a rule-set replaces the fully connected layers—comes mainly from post-training binarization. To fix it, the authors train the CNN with an added sparsity loss that pushes a pre-selected set of class-specific filters toward 1 and all others toward 0, so the outputs are already nearly binary when the rule-set is extracted. On several scene and traffic-sign datasets, the best training strategy raises neuro-symbolic accuracy by 9% on average over the prior state of the art and shrinks the rule-set by 53%, closing the gap to the original CNN to about 3%. This would make interpretable rule-sets a much more viable substitute for black-box CNNs.

What carries the argument

The key object is the sparsity loss function. For each training image, the L2 norm of each last-layer filter is centered by subtracting a precomputed threshold (a per-filter mean plus scaled standard deviation), passed through a sigmoid, and then compared via binary cross-entropy to a target of 1 for the selected class-specific filters and 0 for all others. The selection is encoded in a filter probability matrix $P$ of shape $C \times F$, and the thresholds are computed from the training set before the loss is applied. This setup turns filter outputs into pseudo-binary values, so rounding them to 0/1 after training is nearly lossless. The paper tests five training strategies that vary how $P$ is chosen and when the loss is activated; TS3, which uses a random $P$ and applies the loss from the start alongside cross-entropy, gives the best accuracy and rule-set size.

What would settle it

Train the same VGG16 with the TS3 sparsity loss, then compare rule extraction from the pseudo-binary sigmoid values versus from the raw continuous L2 norms fed to a rule learner that accepts real-valued inputs (e.g., a soft decision tree). If the continuous-pipeline accuracy is still more than 3% below the CNN, then binarization is not the main source of loss. Alternatively, run TS3 but replace FOLD-SE-M with a standard decision-tree extractor on the same binarized vectors; if the 9% gain over EBP disappears, the benefit is specific to FOLD-SE-M rather than to the binarization.

Watch

Extended reading notes

Core claim

The paper's central claim is that training a CNN with a class-specific sparsity loss—one that drives a pre-selected set of K filter outputs per class toward 1 and all other filter outputs toward 0 via binary cross-entropy on sigmoid-adjusted L2 norms—makes the filter activations nearly binary before rule extraction. When these rounded activations are fed to the FOLD-SE-M rule learner, the extracted logic program is both more accurate and much smaller than the one produced by the prior EBP-based pipeline. Surprisingly, the best-performing strategy assigns the per-class filter targets randomly rather than by activation strength; the authors interpret this as showing that the binarization pressure itself, not the choice of which filters to keep, does the work. As a result, the neuro-symbolic model comes within 3% of the original CNN's accuracy on average across the tested datasets.

Load-bearing premise

The load-bearing premise is that the accuracy gap between a CNN and its neuro-symbolic surrogate is caused mainly by post-training binarization of filter activations; if the gap comes instead from the rule-learning algorithm or the threshold choices, the sparsity loss would not address the real problem.

Editorial extensions

If this is right

  • If the loss works as described, neuro-symbolic classifiers can match a fine-tuned CNN to within 3% accuracy, making rule-based explanations viable for deployment in high-stakes image classification.
  • The 53% average reduction in rule-set size means extracted logic programs have far fewer predicates, which lowers the human cost of auditing and verifying the rules.
  • The finding that a randomly assigned filter-class map (TS3) beats activation-based selection (TS2) suggests the method is robust to filter initialization and may not require dataset-specific hyperparameter tuning for the filter map.
  • On 10-class and 43-class datasets, the gains over the baseline increase, indicating the sparsity loss scales with label count rather than degrading.
  • The success of TS4, which uses only the sparsity loss and no cross-entropy, shows that binarized filter features alone can support rule extraction, though with lower fidelity to the original CNN.

Reading between the lines

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

  • We infer that if binarization is indeed the root cause, the sparsity loss should transfer to other rule extractors (e.g., decision trees or rule lists) and other CNN backbones; this is testable and the paper does not run it.
  • The robustness to random filter assignment hints that the loss's regularizing effect, rather than the semantic content of filters, drives the gains; a useful experiment would compare random maps to adversarially chosen maps to see if any fixed partition works.
  • The authors' future-work suggestion of extending the loss to Vision Transformers is natural, but the threshold-and-sigmoid construction relies on spatial feature-map norms, so an adaptation would need a different normalization for attention outputs.
  • In safety-critical settings, the smaller rule-sets combined with answer-set-programming justification could allow formal verification of individual predictions; the paper demonstrates the justification machinery but does not quantify verification cost.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 3 minor

Summary. The paper argues that the accuracy loss incurred by NeSyFOLD-style rule extraction stems from post-training binarization of CNN filter activations. To mitigate this, it proposes a sparsity loss that assigns each class a small set of filters with target activation values of 1 and pushes all other filter outputs toward 0, using a threshold-adjusted sigmoid and a binary cross-entropy term. Five training strategies (TS1--TS5) are evaluated on Places subsets (P2--P10) and GTSRB. The main claim is that TS3, which uses random class--filter assignment, improves NeSy model accuracy by 9% and reduces rule-set size by 53% on average relative to NeSyFOLD-EBP, while coming within 3% of the original CNN's accuracy. The paper also provides qualitative evidence that the learned filters correspond to class-relevant concepts and gives usage guidance for the different strategies.

Significance. If the reported results are robust, the proposed loss is a meaningful step toward making neuro-symbolic image classifiers competitive with black-box CNNs. The method is simple, the experimental setup is clearly described, and the comparison to NeSyFOLD-EBP follows the same pipeline on the same datasets, which is a strength. The claims are also falsifiable through the stated research questions. However, the headline 9% and 53% figures depend on a randomly initialized assignment whose variance is not reported, and the central causal claim that post-training binarization is the root cause of the accuracy gap is not directly tested. These issues currently limit the confidence one can place in the benchmark numbers.

major comments (3)
  1. [Section 4, Table 1 (TS3)] The main accuracy/rule-set-size claim (9% improvement, 53% reduction) uses TS3, where the P matrix is initialized randomly via Method 2 in Section 3.1. The text says results are averaged over 5 runs, but it does not state whether the random assignment is resampled across those runs, and no standard deviation, per-run values, or significance tests are reported for any table. Because the CNN starts from fixed ImageNet weights and the data splits are fixed, the random P matrix is the main uncontrolled source of variation. Without a measure of this variance, the reported means do not establish that the headline numbers are reproducible.
  2. [Section 1 and Section 4 (causal claim)] The paper states that post-training binarization of filter activations is the root cause of the accuracy loss, but no experiment directly tests this claim. TS2 and TS3 reduce the accuracy gap, but this could also result from the sparsity loss acting as a generic feature-separability regularizer rather than from the specific pseudo-binarization mechanism. An ablation is needed: for example, train with the same sparsity loss but replace the binary target assignment with a non-binarizing feature-separation objective, or compare against a baseline that uses the same thresholds but a different binarization scheme. Without such an ablation, the attribution of the gains to 'binarization during training' is underdetermined.
  3. [Section 4 (Hyperparameters)] The method depends on several hyperparameters -- alpha, beta, h1, h2, K, ratio, and tail -- but no sensitivity analysis is reported. The guidance in Section 6 about when to use TS2 versus TS3 versus TS4 is only meaningful if performance is stable under reasonable variations of these settings. At minimum, the authors should report the effect of beta and h1/h2 on one dataset, and of K on one larger dataset, especially because the threshold subtraction in Equations (2)--(5) is shown to be critical by the degraded TS5 results.
minor comments (3)
  1. [Table 2 caption] The caption contains a typo: 'NeSy mdoel' should be 'NeSy model'.
  2. [Section 2 and 4] There are small formatting issues: 'FOLD-SE-M[Wang and Gupta, 2024]' and 'FOLD-SE-M [Wang and Gupta, 2024 ]' are inconsistent, and 'abx' should be typeset consistently as a predicate symbol.
  3. [Section 4, Q4] The qualitative filter-analysis result is based on selecting one top filter per class from the rule-set. The paper does not report how many filters were examined or whether the shown examples are representative; a quantitative concept-alignment metric would strengthen this section.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: the accuracy and rule-size gains are empirical measurements, not consequences of the loss definition by construction.

full rationale

The paper's central claim—that the proposed sparsity loss improves neuro-symbolic rule extraction accuracy and compactness—is established empirically, not derived from the loss definition. The loss L = α·L_cross-entropy + β·L_sparsity is a training objective; the reported +9% accuracy and −53% rule-set size are measured on held-out test sets of Places subsets and GTSRB using the FOLD-SE-M rule extractor. No fitted parameter is renamed as a prediction: the P matrix (Top-K or random) and thresholds h1, h2 are set before training, and the binarized feature vectors are then used to train rules. The baseline NeSyFOLD-EBP is the authors' prior work, so the comparison is same-group, but no load-bearing argument reduces to that citation: the numbers are measured and the rule-extraction algorithm is independently published. The root-cause claim about post-training binarization is asserted rather than directly ablated, and the surprising TS3 random-assignment result lacks variance reporting, but these are empirical robustness concerns, not circularity. Accordingly, no circular step is found.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The central claim rests on several hand-chosen hyperparameters and on assumptions about the binarization proxy and rule extraction. No code or data are provided, so these assumptions cannot be independently checked.

free parameters (7)
  • alpha = 1
    Weight for cross-entropy loss in Eq. (8); chosen by hand, not fitted.
  • beta = 5
    Weight for sparsity loss in Eq. (8); chosen by hand.
  • K = 5 (P2/P3.x), 20 (P10/GT43)
    Number of Top-K filters per class in the P matrix; set per dataset.
  • h1 = 0.6
    Threshold hyperparameter in Eq. (2).
  • h2 = 0.7
    Threshold hyperparameter in Eq. (2).
  • ratio = 0.8
    FOLD-SE-M hyperparameter that limits the false-positive ratio.
  • tail = 0.005
    FOLD-SE-M hyperparameter for minimum rule coverage.
assumptions (4)
  • domain assumption ImageNet-pretrained VGG16 provides a good enough feature basis for the target datasets.
    The method starts from pretrained weights and fine-tunes; if these features are poor, the rule-set accuracy would be poor. See Section 4 Setup.
  • ad hoc to paper The sigmoid of the L2 norm of a filter's feature map, after threshold subtraction, is a faithful indicator of the filter's activation for classification.
    This is the core binarization proxy introduced in Section 3.1; no theoretical justification is given.
  • domain assumption FOLD-SE-M can extract accurate and small rule sets from the binarized feature table.
    The rule extraction is treated as a black box; its behavior is not analyzed. See Section 3.2.
  • ad hoc to paper The random assignment of filters to classes in Method 2 is fixed and does not need to be updated during training.
    TS3 relies on this; if the assignment were unlucky, performance could degrade. See Section 3.1 Method 2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Improving Interpretability and Accuracy in Neuro-Symbolic Rule Extraction Using Class-Specific Sparse Filters." pith.science (2026). https://pith.science/paper/HSGYGTYP

@misc{pith2026250116677,
  author       = {Pith},
  title        = {Pith review of: Improving Interpretability and Accuracy in Neuro-Symbolic Rule Extraction Using Class-Specific Sparse Filters},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HSGYGTYP}},
  note         = {Machine review of arXiv:2501.16677}
}
read the original abstract

There has been significant focus on creating neuro-symbolic models for interpretable image classification using Convolutional Neural Networks (CNNs). These methods aim to replace the CNN with a neuro-symbolic model consisting of the CNN, which is used as a feature extractor, and an interpretable rule-set extracted from the CNN itself. While these approaches provide interpretability through the extracted rule-set, they often compromise accuracy compared to the original CNN model. In this paper, we identify the root cause of this accuracy loss as the post-training binarization of filter activations to extract the rule-set. To address this, we propose a novel sparsity loss function that enables class-specific filter binarization during CNN training, thus minimizing information loss when extracting the rule-set. We evaluate several training strategies with our novel sparsity loss, analyzing their effectiveness and providing guidance on their appropriate use. Notably, we set a new benchmark, achieving a 9% improvement in accuracy and a 53% reduction in rule-set size on average, compared to the previous SOTA, while coming within 3% of the original CNN's accuracy. This highlights the significant potential of interpretable neuro-symbolic models as viable alternatives to black-box CNNs.

Figures

Figures reproduced from arXiv: 2501.16677 by the authors.

Figure 1
Figure 1. The NeSyFOLD Framework Specifically, in the context of image classification using CNNs, there has been a lot of effort towards improving their interpretability by extracting a rule-set from the convolution layers, that explains the underlying decision-making logic of the model [Townsend et al., 2021; Townsend et al., 2022; Padalkar et al., 2024a; Padalkar et al., 2024b; Padalkar et al., 2024c]. The rule-set along wi… view at source ↗
Figure 2
Figure 2. Activation maps of the top filters for the training strategies [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Raw rule-set produced via TS3 for P3.1 dataset (top). La [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 37 canonical work pages

  1. [1]

    Robert Andrews, Joachim Diederich, and Alan B. Tickle. Survey and critique of techniques for extracting rules from trained artificial neural networks. Knowledge-Based Systems , 8(6):373--389, 1995. Knowledge-based neural networks

    work page 1995

  2. [2]

    Constraint answer set programming without grounding

    Joaqu \'i n Arias, Manuel Carro, Elmer Salazar, Kyle Marple, and Gopal Gupta. Constraint answer set programming without grounding. Theory and Practice of Logic Programming , 18(3-4):337--354, 2018

    work page 2018

  3. [3]

    Knowledge representation, reasoning and declarative problem solving

    Chitta Baral. Knowledge representation, reasoning and declarative problem solving . Cambridge University Press, 2003

    work page 2003

  4. [4]

    Sparse autoencoders find highly interpretable features in language models

    Hoagy Cunningham, Aidan Ewart, Logan Riggs, Robert Huben, and Lee Sharkey. Sparse autoencoders find highly interpretable features in language models. arXiv preprint arXiv:2309.08600 , 2023

    arXiv 2023

  5. [5]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In Proc. CVPR , pages 248--255. Ieee, 2009

    work page 2009

  6. [6]

    Extraction of salient sentences from labelled documents

    Misha Denil, Alban Demiraj, and Nando de Freitas. Extraction of salient sentences from labelled documents. ArXiv , abs/1412.6815, 2014

    arXiv 2014

  7. [7]

    Accelerating convolutional neural networks via activation map compression

    Georgios Georgiadis. Accelerating convolutional neural networks via activation map compression. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 7085--7095, 2019

    work page 2019

  8. [8]

    Ozan Irsoy, O. T. Yildiz, and Ethem Alpaydin. Soft decision trees. Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012) , pages 1819--1822, 2012

    work page 2012

Show all 41 references
  1. [9]

    Deep learning using computer vision in self driving cars for lane and traffic sign detection

    Nitin Kanagaraj, David Hicks, Ayush Goyal, Sanju Mishra Tiwari, and Ghanapriya Singh. Deep learning using computer vision in self driving cars for lane and traffic sign detection. International Journal of System Assurance Engineering and Management , 12, 05 2021

    2021

  2. [10]

    Elite backprop: Training sparse interpretable neurons

    Theodoros Kasioumis, Joe Townsend, and Hiroya Inakoshi. Elite backprop: Training sparse interpretable neurons. In International Workshop on Neural-Symbolic Learning and Reasoning , 2021

    2021

  3. [11]

    Sparseout: Controlling sparsity in deep networks

    Najeeb Khan and Ian Stavness. Sparseout: Controlling sparsity in deep networks. In Canadian conference on artificial intelligence , pages 296--307. Springer, 2019

    2019

  4. [12]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. Adam: A method for stochastic optimization. In Proc. ICLR , 2015

    2015

  5. [13]

    Survey of computer vision-based natural disaster warning systems

    Byoungchul Ko and Sooyeong Kwak. Survey of computer vision-based natural disaster warning systems. Optical Engineering , 51:0901--, 07 2012

    2012

  6. [14]

    Inducing and exploiting activation sparsity for fast inference on deep neural networks

    Mark Kurtz, Justin Kopinsky, Rati Gelashvili, Alexander Matveev, John Carr, Michael Goin, William Leiserson, Sage Moore, Nir Shavit, and Dan Alistarh. Inducing and exploiting activation sparsity for fast inference on deep neural networks. In International Conference on Machine...

    2020

  7. [15]

    Human evaluation of models built for interpretability

    Isaac Lage, Emily Chen, Jeffrey He, Menaka Narayanan, Been Kim, Samuel J Gershman, and Finale Doshi-Velez. Human evaluation of models built for interpretability. In Proc. HCOMP , volume 7, pages 59--67, 2019

    2019

  8. [16]

    Boser, John S

    Yann LeCun, Bernhard E. Boser, John S. Denker, Donnie Henderson, Richard E. Howard, Wayne E. Hubbard, and Lawrence D. Jackel. Backpropagation applied to handwritten zip code recognition. Neural Comput. , 1(4):541--551, 1989

    1989

  9. [17]

    Interpretable deep learning: Interpretation, interpretability, trustworthiness, and beyond

    Xuhong Li, Haoyi Xiong, Xingjian Li, Xuanyu Wu, Xiao Zhang, Ji Liu, Jiang Bian, and Dejing Dou. Interpretable deep learning: Interpretation, interpretability, trustworthiness, and beyond. Knowledge and Information Systems , 64(12):3197--3234, 2022

    2022

  10. [18]

    Training interpretable convolutional neural networks by differentiating class-specific filters

    Haoyu Liang, Zhihao Ouyang, Yuyuan Zeng, Hang Su, Zihao He, Shu-Tao Xia, Jun Zhu, and Bo Zhang. Training interpretable convolutional neural networks by differentiating class-specific filters. In Computer Vision – ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, ...

    2020

  11. [19]

    A survey of sparse-learning methods for deep neural networks

    Rongrong Ma and Lingfeng Niu. A survey of sparse-learning methods for deep neural networks. In 2018 IEEE/WIC/ACM International Conference on Web Intelligence (WI) , pages 647--650. IEEE, 2018

    2018

  12. [20]

    Extracting meaningful high-fidelity knowledge from convolutional neural networks

    Kwun Ho Ngan, Artur D'Avila Garcez, and Joseph Townsend. Extracting meaningful high-fidelity knowledge from convolutional neural networks. In 2022 International Joint Conference on Neural Networks (IJCNN) , pages 1--17, 2022

    2022

  13. [21]

    Nesyfold: A framework for interpretable image classification

    Parth Padalkar, Huaduo Wang, and Gopal Gupta. Nesyfold: A framework for interpretable image classification. In Proc. AAAI , pages 4378--4387. AAAI Press, 2024

    2024

  14. [22]

    Using logic programming and kernel-grouping for improving interpretability of convolutional neural networks

    Parth Padalkar, Huaduo Wang, and Gopal Gupta. Using logic programming and kernel-grouping for improving interpretability of convolutional neural networks. In Proc. PADL , volume 14512 of LNCS , pages 134--150. Springer, 2024

    2024

  15. [23]

    A neurosymbolic framework for bias correction in convolutional neural networks

    Parth Padalkar, Natalia Ślusarz, Ekaterina Komendantskaya, and Gopal Gupta. A neurosymbolic framework for bias correction in convolutional neural networks. Theory and Practice of Logic Programming , 24(4):644–662, 2024

    2024

  16. [24]

    Toward transparent ai: A survey on interpreting the inner structures of deep neural networks

    Tilman R \"a uker, Anson Ho, Stephen Casper, and Dylan Hadfield-Menell. Toward transparent ai: A survey on interpreting the inner structures of deep neural networks. In 2023 ieee conference on secure and trustworthy machine learning (satml) , pages 464--483. IEEE, 2023

    2023

  17. [25]

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

    Ramprasaath R Selvaraju, Michael Cogswell, Abhishek Das, Ramakrishna Vedantam, Devi Parikh, and Dhruv Batra. Grad-cam: Visual explanations from deep networks via gradient-based localization. In Proceedings of the IEEE international conference on computer vision , pages 618--626, 2017

    2017

  18. [26]

    Interpretable compositional convolutional neural networks

    Wen Shen, Zhihua Wei, Shikun Huang, Binbin Zhang, Jiaqi Fan, Ping Zhao, and Quanshi Zhang. Interpretable compositional convolutional neural networks. arXiv preprint arXiv:2107.04474 , 2021

    2021 arXiv

  19. [27]

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

    Karen Simonyan, Andrea Vedaldi, and Andrew Zisserman. Deep inside convolutional networks: Visualising image classification models and saliency maps. CoRR , abs/1312.6034, 2013

    2013 arXiv

  20. [28]

    Dropout: a simple way to prevent neural networks from overfitting

    Nitish Srivastava, Geoffrey Hinton, Alex Krizhevsky, Ilya Sutskever, and Ruslan Salakhutdinov. Dropout: a simple way to prevent neural networks from overfitting. The journal of machine learning research , 15(1):1929--1958, 2014

    1929

  21. [29]

    Stallkamp, M

    J. Stallkamp, M. Schlipsing, J. Salmen, and C. Igel. Man vs. computer: Benchmarking machine learning algorithms for traffic sign recognition. Neural Networks , 32:323--332, 2012. Selected Papers from IJCNN 2011

    2012

  22. [30]

    Sun, Bin Zheng, and Wei Qian

    W. Sun, Bin Zheng, and Wei Qian. Computer aided lung cancer diagnosis with deep learning algorithms. In SPIE Medical Imaging , 2016

    2016

  23. [31]

    Tickle, R

    A.B. Tickle, R. Andrews, M. Golea, and J. Diederich. The truth will come to light: directions and challenges in extracting the knowledge embedded within trained artificial neural networks. IEEE Transactions on Neural Networks , 9(6):1057--1068, 1998

    1998

  24. [32]

    Eric: Extracting relations inferred from convolutions

    Joe Townsend, Theodoros Kasioumis, and Hiroya Inakoshi. Eric: Extracting relations inferred from convolutions. In Computer Vision -- ACCV , pages 206--222, Cham, 2021. Springer International

    2021

  25. [33]

    On the explainability of convolutional layers for multi-class problems

    Joe Townsend, Mateusz Kudla, Agnieszka Raszkowska, and Theodoros Kasiousmis. On the explainability of convolutional layers for multi-class problems. In Combining Learning and Reasoning: Programming Languages, Formalisms, and Representations , 2022

    2022

  26. [34]

    FOLD-SE: an efficient rule-based machine learning algorithm with scalable explainability

    Huaduo Wang and Gopal Gupta. FOLD-SE: an efficient rule-based machine learning algorithm with scalable explainability. In Proc. PADL , volume 14512 of LNCS , pages 37--53. Springer, 2024

    2024

  27. [35]

    Dasnet: Dynamic activation sparsity for neural network efficiency improvement

    Qing Yang, Jiachen Mao, Zuoguan Wang, and Hai Li. Dasnet: Dynamic activation sparsity for neural network efficiency improvement. In 2019 IEEE 31st International Conference on Tools with Artificial Intelligence (ICTAI) , pages 1401--1405. IEEE, 2019

    2019

  28. [36]

    Visualizing and understanding convolutional networks

    Matthew D Zeiler and Rob Fergus. Visualizing and understanding convolutional networks. In European conference on computer vision , pages 818--833. Springer, 2014

    2014

  29. [37]

    Interpretable convolutional neural networks

    Quanshi Zhang, Ying Nian Wu, and Song-Chun Zhu. Interpretable convolutional neural networks. In Proceedings of the IEEE conference on computer vision and pattern recognition , pages 8827--8836, 2018

    2018

  30. [38]

    A survey on neural network interpretability

    Yu Zhang, Peter Ti n o, Ale s Leonardis, and Ke Tang. A survey on neural network interpretability. IEEE Transactions on Emerging Topics in Computational Intelligence , 5(5):726--742, 2021

    2021

  31. [39]

    Object detectors emerge in deep scene cnns

    Bolei Zhou, Aditya Khosla, \` A gata Lapedriza, Aude Oliva, and Antonio Torralba. Object detectors emerge in deep scene cnns. In ICLR Conference Track Proceedings , 2015

    2015

  32. [40]

    Places: A 10 million image database for scene recognition, 2017

    Bolei Zhou, Agata Lapedriza, Aditya Khosla, Aude Oliva, and Antonio Torralba. Places: A 10 million image database for scene recognition, 2017

    2017

  33. [41]

    write newline

    " write newline "" before.all 'output.state := FUNCTION fin.entry add.period write newline FUNCTION new.block output.state before.all = 'skip after.block 'output.state := if FUNCTION new.sentence output.state after.block = 'skip output.state before.all = 'skip after.sentence '...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.