Explainable Neural Networks

The Reliability of Deep Learning for Signal and Image Processing: Interpretability, Robustness, and Accuracy
This thesis investigates several aspects of using data-driven methods for image and signal processing tasks, particularly those aspects …
Jan Macdonald
PDF Cite DOI
Interpretable Neural Networks with Frank-Wolfe: Sparse Relevance Maps and Relevance Orderings
We study the effects of constrained optimization formulations and Frank-Wolfe algorithms for obtaining interpretable neural network …
Jan Macdonald, Mathieu Besançon, Sebastian Pokutta
Preprint PDF Cite Code
Interpretable Neural Networks with Frank-Wolfe: Sparse Relevance Maps and Relevance Orderings
Spotlight & Poster: Presentation of results from our paper.
Jul 20, 2022 10:30 AM — 10:35 AM Baltimore Convention Center, Baltimore, USA
Jan Macdonald, Mathieu Besançon, Sebastian Pokutta
Code Video
The Computational Complexity of Understanding Binary Classifier Decisions
For a $d$-ary Boolean function $\Phi\colon\{0,1\}^d\to\{0,1\}$ and an assignment to its variables $\mathbf{x}=(x_1, x_2, \dots, x_d)$ …
Stephan Wäldchen, Jan Macdonald, Sascha Hauch, Gitta Kutyniok
Preprint PDF Cite DOI
Explaining Neural Network Decisions Is Hard
Poster: Presentation of results based on our papers A Rate-Distortion Framework for Explaining Neural Network Decisions and The Computational Complexity of Understanding Network Decisions.
Jul 17, 2020 Virtual Workshop
Jan Macdonald, Stephan Wäldchen, Sascha Hauch, Gitta Kutyniok
Explaining Neural Network Decisions Is Hard
We connect the widespread idea of interpreting classifier decisions to probabilistic prime implicants. A set of input features is deemed …
Jan Macdonald, Stephan Wäldchen, Sascha Hauch, Gitta Kutyniok
PDF Cite Code
A Rate-Distortion Framework for Explaining Neural Network Decisions
Poster: Presentation of results from our paper.
Oct 24, 2019 Zuse Institut Berlin
Jan Macdonald, Stephan Wäldchen, Sascha Hauch, Gitta Kutyniok
A Rate-Distortion Framework for Explaining Neural Network Decisions
Poster: Presentation of results from our paper.
Sep 9, 2019 Fraunhofer Heinrich-Hertz-Institut
Jan Macdonald, Stephan Wäldchen, Sascha Hauch, Gitta Kutyniok
A Rate-Distortion Framework for Explaining Deep Neural Network Decisions
Invited Talk: Presentation of results from our paper.
Jul 17, 2019 5:30 PM — 6:00 PM Universitat de València
Jan Macdonald, Stephan Wäldchen, Sascha Hauch, Gitta Kutyniok
A Rate-Distortion Framework for Explaining Neural Network Decisions
Poster: Presentation of results from our paper.
Jul 3, 2019 École nationale supérieure d'électrotechnique, d'électronique, d'informatique, d'hydraulique et des télécommunications (ENSEEIHT)
Jan Macdonald, Stephan Wäldchen, Sascha Hauch, Gitta Kutyniok