ManiFool
Algorithm for evaluating the invariance properties of deep networks
Deep convolutional neural networks have been shown to be vulnerable to arbitrary geometric transformations. However, there is no systematic method to measure the invariance properties of deep networks to such transformations. ManiFool is a simple yet scalable algorithm to measure the invariance of deep networks. In particular, it measures the robustness of deep networks to geometric transformations in a worst-case regime as they can be problematic for sensitive applications.
Deep Neural Networks
- Technical
- Research papers
- Source code: Personal GitHub
- Last commit: 2018-01-24
The Signal Processing Laboratory (LTS4) is a team of researchers led by Prof. Pascal Frossard, working in the Electrical Engineering Institute of the Swiss Federal Institute of Technology (EPFL), Lausanne, Switzerland.
The group research focuses on image processing, graph signal processing and machine learning, as well as closely related fields such as network data analysis, distributed signal processing, image and video coding and immersive communications. We work at the frontier between signal processing, machine learning and applied mathematics.
The group research focuses on image processing, graph signal processing and machine learning, as well as closely related fields such as network data analysis, distributed signal processing, image and video coding and immersive communications. We work at the frontier between signal processing, machine learning and applied mathematics.
This page was last edited on 2024-03-21.
This page was last edited on 2024-03-21.