Zobrazeno 1 - 7
of 7
pro vyhledávání: '"Lucas, Mirtha"'
Autor:
Lerma, Miguel, Lucas, Mirtha
We discuss a vulnerability involving a category of attribution methods used to provide explanations for the outputs of convolutional neural networks working as classifiers. It is known that this type of networks are vulnerable to adversarial attacks,
Externí odkaz:
http://arxiv.org/abs/2307.03305
Autor:
Lerma, Miguel, Lucas, Mirtha
Publikováno v:
ICPRS 13 (2023) 1-4
Gradient based attribution methods for neural networks working as classifiers use gradients of network scores. Here we discuss the practical differences between using gradients of pre-softmax scores versus post-softmax scores, and their respective ad
Externí odkaz:
http://arxiv.org/abs/2306.13197
Neural networks are becoming increasingly better at tasks that involve classifying and recognizing images. At the same time techniques intended to explain the network output have been proposed. One such technique is the Gradient-based Class Activatio
Externí odkaz:
http://arxiv.org/abs/2205.10900
Autor:
Lerma, Miguel, Lucas, Mirtha
The Grad-CAM algorithm provides a way to identify what parts of an image contribute most to the output of a classifier deep network. The algorithm is simple and widely used for localization of objects in an image, although some researchers have point
Externí odkaz:
http://arxiv.org/abs/2205.10838
Autor:
Lerma, Miguel, Lucas, Mirtha
We discuss a way to find a well behaved baseline for attribution methods that work by feeding a neural network with a sequence of interpolated inputs between two given inputs. Then, we test it with our novel Riemann-Stieltjes Integrated Gradient-weig
Externí odkaz:
http://arxiv.org/abs/2204.06120
Autor:
Lerma, Miguel, Lucas, Mirtha
We provide rigorous proofs that the Integrated Gradients (IG) attribution method for deep networks satisfies completeness and symmetry-preserving properties. We also study the uniqueness of IG as a path method preserving symmetry.
Comment: 7 pag
Comment: 7 pag
Externí odkaz:
http://arxiv.org/abs/2103.13533
Publikováno v:
EAI Endorsed Transactions on Bioengineering & Bioinformatics; 2021, Vol. 1 Issue 2, p1-12, 12p
