Tropical Decision Boundaries for Neural Networks Are Robust Against Adversarial Attacks
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916112135880704 |
|---|---|
| author | Pasque, Kurt Teska, Christopher Yoshida, Ruriko Miura, Keiji Huang, Jefferson |
| author_facet | Pasque, Kurt Teska, Christopher Yoshida, Ruriko Miura, Keiji Huang, Jefferson |
| contents | We introduce a simple, easy to implement, and computationally efficient tropical convolutional neural network architecture that is robust against adversarial attacks. We exploit the tropical nature of piece-wise linear neural networks by embedding the data in the tropical projective torus in a single hidden layer which can be added to any model. We study the geometry of its decision boundary theoretically and show its robustness against adversarial attacks on image datasets using computational experiments. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_00576 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Tropical Decision Boundaries for Neural Networks Are Robust Against Adversarial Attacks Pasque, Kurt Teska, Christopher Yoshida, Ruriko Miura, Keiji Huang, Jefferson Machine Learning Cryptography and Security Computer Vision and Pattern Recognition Combinatorics We introduce a simple, easy to implement, and computationally efficient tropical convolutional neural network architecture that is robust against adversarial attacks. We exploit the tropical nature of piece-wise linear neural networks by embedding the data in the tropical projective torus in a single hidden layer which can be added to any model. We study the geometry of its decision boundary theoretically and show its robustness against adversarial attacks on image datasets using computational experiments. |
| title | Tropical Decision Boundaries for Neural Networks Are Robust Against Adversarial Attacks |
| topic | Machine Learning Cryptography and Security Computer Vision and Pattern Recognition Combinatorics |
| url | https://arxiv.org/abs/2402.00576 |