B-cos Alignment for Inherently Interpretable CNNs and Vision Transformers

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Böhle, Moritz, Singh, Navdeeppal, Fritz, Mario, Schiele, Bernt
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909072525099008
author Böhle, Moritz
Singh, Navdeeppal
Fritz, Mario
Schiele, Bernt
author_facet Böhle, Moritz
Singh, Navdeeppal
Fritz, Mario
Schiele, Bernt
contents 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 transformations in DNNs by our novel B-cos transformation. As we show, a sequence (network) of such transformations induces a single linear transformation that faithfully summarises the full model computations. Moreover, the B-cos transformation is designed such that the weights align with relevant signals during optimisation. As a result, those induced linear transformations become highly interpretable and highlight task-relevant features. Importantly, the B-cos transformation is designed to be compatible with existing architectures and we show that it can easily be integrated into virtually all of the latest state of the art models for computer vision - e.g. ResNets, DenseNets, ConvNext models, as well as Vision Transformers - by combining the B-cos-based explanations with normalisation and attention layers, all whilst maintaining similar accuracy on ImageNet. Finally, we show that the resulting explanations are of high visual quality and perform well under quantitative interpretability metrics.
format Preprint
id arxiv_https___arxiv_org_abs_2306_10898
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle B-cos Alignment for Inherently Interpretable CNNs and Vision Transformers
Böhle, Moritz
Singh, Navdeeppal
Fritz, Mario
Schiele, Bernt
Computer Vision and Pattern Recognition
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 transformations in DNNs by our novel B-cos transformation. As we show, a sequence (network) of such transformations induces a single linear transformation that faithfully summarises the full model computations. Moreover, the B-cos transformation is designed such that the weights align with relevant signals during optimisation. As a result, those induced linear transformations become highly interpretable and highlight task-relevant features. Importantly, the B-cos transformation is designed to be compatible with existing architectures and we show that it can easily be integrated into virtually all of the latest state of the art models for computer vision - e.g. ResNets, DenseNets, ConvNext models, as well as Vision Transformers - by combining the B-cos-based explanations with normalisation and attention layers, all whilst maintaining similar accuracy on ImageNet. Finally, we show that the resulting explanations are of high visual quality and perform well under quantitative interpretability metrics.
title B-cos Alignment for Inherently Interpretable CNNs and Vision Transformers
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2306.10898