Explainable Image Classification with Evidence Counterfactual

Vermeire, Tom; Martens, David

doi:10.48550/arxiv.2004.07511

Cited by 8 publications

(10 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Also note that both SEDC and CFproto are not limited to tabular data. For other applications such as explaining images, these algorithms can also be useful [40,22].…”

Section: Explanation Requirementsmentioning

confidence: 99%

NICE: An Algorithm for Nearest Instance Counterfactual Explanations

Brughmans¹,

Leyman²,

Martens³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

In this paper we suggest NICE: a new algorithm to generate counterfactual explanations for heterogeneous tabular data. The design of our algorithm specifically takes into account algorithmic requirements that often emerge in real-life deployments: the ability to provide an explanation for all predictions, being efficient in run-time, and being able to handle any classification model (also nondifferentiable ones). More specifically, our approach exploits information from a nearest instance to speed up the search process. We propose four versions of NICE, where three of them optimize the explanations for one of the following properties: sparsity, proximity or plausibility. An extensive empirical comparison on 10 datasets shows that our algorithm performs better on all properties than the current state-of-the-art. These analyses show a trade-off between on the one hand plausiblity and on the other hand proximity or sparsity, with our different optimization methods offering the choice to select the preferred trade-off. An open-source implementation of NICE can be found at https://github.com/ADMAntwerp/NICE.

show abstract

“…Also note that both SEDC and CFproto are not limited to tabular data. For other applications such as explaining images, these algorithms can also be useful [40,22].…”

Section: Explanation Requirementsmentioning

confidence: 99%

NICE: An Algorithm for Nearest Instance Counterfactual Explanations

Brughmans¹,

Leyman²,

Martens³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Counterfactuals provide explanations for chosen decisions by describing what changes on the input would lead to an alternative prediction while minimizing the magnitude of the changes to preserve the fidelity, which is identical to the process of generating adversarial examples [6]. Unfortunately, owing to the multidimensional geometric information that is unacceptable to the human brain, existing image-oriented approaches addressed the counterfactual explanations only at the semantic level [11,32].…”

Section: Towards Explainable Pc Modelsmentioning

confidence: 99%

Explainability-Aware One Point Attack for Point Cloud Neural Networks

Tan¹,

Kotthaus²

2021

Preprint

View full text Add to dashboard Cite

With the proposition of neural networks for point clouds, deep learning has started to shine in the field of 3D object recognition while researchers have shown an increased interest to investigate the reliability of point cloud networks by fooling them with perturbed instances. However, most studies focus on the imperceptibility or surface consistency, with humans perceiving no perturbations on the adversarial examples. This work proposes two new attack methods: one-point attack (OPA) and critical traverse attack (CTA), which go in the opposite direction: we restrict the perturbation dimensions to a human cognizable range with the help of explainability methods, which enables the working principle or decision boundary of the models to be comprehensible through the observable perturbation magnitude. Our results show that the popular point cloud networks can be deceived with almost 100% success rate by shifting only one point from the input instance. In addition, we attempt to provide a more persuasive viewpoint of comparing the robustness of point cloud models against adversarial attacks. We also show the interesting impact of different point attribution distributions on the adversarial robustness of point cloud networks. Finally, we discuss how our approaches facilitate the explainability study for point cloud networks. To the best of our knowledge, this is the first point-cloud-based adversarial approach concerning explainability. Our code is available at https://github.com/Explain3D/ Exp-One-Point-Atk-PC. Figure 1: One point attack for point cloud networks. With the saliency map provided by the explainability method, only one point needs to be perturbed in the point set of the original instance to fool the most popular point cloud networks.

show abstract

“…For example, lips, eyes, and maybe cheeks would change if makeup was applied to a face but not hair color or the background. Recently, many methods for generating counterfactual examples have been proposed [1,6,7,10,11,13,23,36,37,40]. A common drawback for all the methods is that they need to query the model under consideration many times.…”

Section: Introductionmentioning

confidence: 99%

ECINN: Efficient Counterfactuals from Invertible Neural Networks

Hvilshøj,

Iosifidis,

Assent

2021

Preprint

View full text Add to dashboard Cite

Counterfactual examples identify how inputs can be altered to change the predicted class of a classifier, thus opening up the black-box nature of, e.g., deep neural networks. We propose a method, ECINN, that utilizes the generative capacities of invertible neural networks for image classification to generate counterfactual examples efficiently. In contrast to competing methods that sometimes need a thousand evaluations or more of the classifier, ECINN has a closed-form expression and generates a counterfactual in the time of only two evaluations. Arguably, the main challenge of generating counterfactual examples is to alter only input features that affect the predicted outcome, i.e., class-dependent features. Our experiments demonstrate how ECINN alters class-dependent image regions to change the perceptual and predicted class of the counterfactuals. Additionally, we extend ECINN to also produce heatmaps (ECINNh) for easy inspection of, e.g., pairwise class-dependent changes in the generated counterfactual examples. Experimentally, we find that ECINNh outperforms established methods that generate heatmap-based explanations.

show abstract

Explainable Image Classification with Evidence Counterfactual

Cited by 8 publications

References 0 publications

NICE: An Algorithm for Nearest Instance Counterfactual Explanations

NICE: An Algorithm for Nearest Instance Counterfactual Explanations

Explainability-Aware One Point Attack for Point Cloud Neural Networks

ECINN: Efficient Counterfactuals from Invertible Neural Networks

Contact Info

Product

Resources

About