In Depth Bayesian Semantic Scene Completion

Gillsjö, David; Åström, Kalle

doi:10.1109/icpr48806.2021.9412403

Cited by 3 publications

(1 citation statement)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The Cauchy-prior model is unreliable as it produced refined adversarial patch predictions for some images but completely overestimated the size of the adversarial patch by classifying large portions of the image as part of an adversarial patch. This is in contrast to the findings by[25] which found that Cauchy-prior improved their model performance. Nevertheless, our experiment outcomes are in line with[26] which found that gaussian priors were suitable, albeit a multivariate gaussian, for 2D convolution kernels.Selection of Bayesian layers.…”

contrasting

confidence: 99%

Detection of Uncertainty in Exceedance of Threshold (DUET): An Adversarial Patch Localizer

Chua¹,

Yu²,

Zhao³

2023

Preprint

View full text Add to dashboard Cite

Development of defenses against physical world attacks such as adversarial patches is gaining traction within the research community. We contribute to the field of adversarial patch detection by introducing an uncertainty-based adversarial patch localizer which localizes adversarial patch on an image, permitting post-processing patch-avoidance or patch-reconstruction. We quantify our prediction uncertainties with the development of Detection of Uncertainties in the Exceedance of Threshold (DUET) algorithm. This algorithm provides a framework to ascertain confidence in the adversarial patch localization, which is essential for safety-sensitive applications such as self-driving cars and medical imaging. We conducted experiments on localizing adversarial patches and found our proposed DUET model outperforms baseline models. We then conduct further analyses on our choice of model priors and the adoption of Bayesian Neural Networks in different layers within our model architecture. We found that isometric gaussian priors in Bayesian Neural Networks are suitable for patch localization tasks and the presence of Bayesian layers in the earlier neural network blocks facilitates top-end localization performance, while Bayesian layers added in the later neural network blocks contribute to better model generalization. We then propose two different well-performing models to tackle different use cases.

show abstract

contrasting

confidence: 99%