B-cos Networks: Alignment is All We Need for Interpretability

• Published in: Proceedings (IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Online) pp. 10319 - 10328
• Main Authors: Bohle, Moritz, Fritz, Mario, Schiele, Bernt
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2022
• Subjects: accountability; Computational modeling; Computer vision; Deep learning; Explainable computer vision; Computer vision theory; Deep learning architectures and techniques; Machine learning; Transparency; fairness; Neural networks; privacy and ethics in vision; Visual reasoning; Training; Transforms; Visualization
• ISSN: 1063-6919
• DOI: 10.1109/CVPR52688.2022.01008

We present a new direction for increasing the interpretability of deep neural networks (DNNs) by promoting weight-input alignment during training. For this, we propose to replace the linear transforms in DNNs by our B-cos transform. As we show, a sequence (network) of such transforms induces a single linear transform that faithfully summarises the full model computations. Moreover, the B-cos transform introduces alignment pressure on the weights during optimisation. As a result, those induced linear transforms become highly interpretable and align with task-relevant features. Importantly, the B-cos transform is designed to be compatible with existing architectures and we show that it can easily be integrated into common models such as VGGs, ResNets, InceptionNets, and DenseNets, whilst maintaining similar performance on ImageNet. The resulting explanations are of high visual quality and perform well under quantitative metrics for interpretability. Code available at github.com/moboehle/B-cos.