A Generic Visualization Approach for Convolutional Neural Networks

Taha, Ahmed; Yang, Xitong; Shrivastava, Abhinav; Davis, Larry S.

doi:10.1007/978-3-030-58520-4_43

Cited by 7 publications

(3 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…saturation, zero-gradient image regions, and false confidence in the output score phenomena [70]) the computational cost Grad-CAMs is negligible compared to other methods that require multiple network forward-passes per image [70,130]. Moreover, in most recent works, Grad-CAM is used as the baseline methods from improvements margins [130][131][132][133][134].…”

Section: Visual Explanationmentioning

confidence: 99%

Deep fair models for complex data: Graphs labeling and explainable face recognition

et al. 2022

View full text Add to dashboard Cite

Section: Visual Explanationmentioning

confidence: 99%

Deep fair models for complex data: Graphs labeling and explainable face recognition

et al. 2022

View full text Add to dashboard Cite

“…Although gradient-based methods might not be the optimal solution for visual explanation (e.g., saturation, zero-gradient image regions, and false confidence in the output score phenomena [156]), the computational cost of Grad-CAMs is negligible when compared to other methods that require multiple network forward-passes per image [156,157]. Moreover, Grad-CAM is considered the reference method in several recent works [157][158][159][160][161].…”

Section: Making the Model Interpretablementioning

confidence: 99%

Toward Learning Trustworthily from Data Combining Privacy, Fairness, and Explainability: An Application to Face Recognition

Franco

Oneto

Navarin

et al. 2021

Entropy

View full text Add to dashboard Cite

In many decision-making scenarios, ranging from recreational activities to healthcare and policing, the use of artificial intelligence coupled with the ability to learn from historical data is becoming ubiquitous. This widespread adoption of automated systems is accompanied by the increasing concerns regarding their ethical implications. Fundamental rights, such as the ones that require the preservation of privacy, do not discriminate based on sensible attributes (e.g., gender, ethnicity, political/sexual orientation), or require one to provide an explanation for a decision, are daily undermined by the use of increasingly complex and less understandable yet more accurate learning algorithms. For this purpose, in this work, we work toward the development of systems able to ensure trustworthiness by delivering privacy, fairness, and explainability by design. In particular, we show that it is possible to simultaneously learn from data while preserving the privacy of the individuals thanks to the use of Homomorphic Encryption, ensuring fairness by learning a fair representation from the data, and ensuring explainable decisions with local and global explanations without compromising the accuracy of the final models. We test our approach on a widespread but still controversial application, namely face recognition, using the recent FairFace dataset to prove the validity of our approach.

show abstract

“…However, this subsymbolism (also known as the opaque or black-box model) is vulnerable to the underlying barrier of explainability in response to critical questions like how a particular trained model arrives at a decision, how certain it is about its decision, if and when it can be trusted, why it makes certain mistakes, and in which part of the learning algorithm or parametric space correction should take place [28], [4]. Explainability in CNNs is linked to post-hoc explainability [18] and, as proposed by Arrieta et al [4], relies on model simplification [56], [36], [23], feature-relevance estimation [6], [33], [29], [38], visualisation [53], [30], [26], [39], [48], [22], and architectural modification [27], [15], [40] to convert a non-interpretable model into an explainable one.…”

Section: Introductionmentioning

confidence: 99%

ADVISE: ADaptive Feature Relevance and VISual Explanations for Convolutional Neural Networks

Dehshibi¹,

Ashtari-Majlan²,

Adhane³

et al. 2022

Preprint

View full text Add to dashboard Cite

To equip Convolutional Neural Networks (CNNs) with explainability, it is essential to interpret how opaque models take specific decisions, understand what causes the errors, improve the architecture design, and identify unethical biases in the classifiers. This paper introduces ADVISE, a new explainability method that quantifies and leverages the relevance of each unit of the feature map to provide better visual explanations. To this end, we propose using adaptive bandwidth kernel density estimation to assign a relevance score to each unit of the feature map with respect to the predicted class. We also propose an evaluation protocol to quantitatively assess the visual explainability of CNN models. We extensively evaluate our idea in the image classification task using AlexNet, VGG16, ResNet50, and Xception pretrained on ImageNet. We compare ADVISE with the state-of-the-art visual explainable methods and show that the proposed method outperforms competing approaches in quantifying feature-relevance and visual explainability while maintaining competitive time complexity. Our experiments further show that ADVISE fulfils the sensitivity and implementation independence axioms while passing the sanity checks. The implementation is accessible for reproducibility purposes on https://github.com/dehshibi/ADVISE.

show abstract

A Generic Visualization Approach for Convolutional Neural Networks

Cited by 7 publications

References 32 publications

Deep fair models for complex data: Graphs labeling and explainable face recognition

Deep fair models for complex data: Graphs labeling and explainable face recognition

Toward Learning Trustworthily from Data Combining Privacy, Fairness, and Explainability: An Application to Face Recognition

ADVISE: ADaptive Feature Relevance and VISual Explanations for Convolutional Neural Networks

Contact Info

Product

Resources

About