Feature Accentuation: Explaining 'what' features respond to in natural images

Christopher J Hamblin, Thomas FEL, Srijani Saha, Talia Konkle, George A. Alvarez

Primary Area: visualization or interpretation of learned representations

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Keywords: Feature Visualization, Attribution, Explainability

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

TL;DR: We develop a technique for augmenting images to reveal 'what' features like about them, which is often unintuitive using attribution methods or feature visualization alone.

Abstract: Efforts to decode neural network vision models necessitate a comprehensive grasp of both the spatial and semantic facets governing feature responses within images. Most research has primarily centered around attribution methods, which provide explanations in the form of heatmaps, showing 'where' the model directs its attention for a given feature. However, grasping 'where' alone falls short, as numerous studies have highlighted the limitations of those methods and the necessity to understand 'what' the model has recognized at the focal point of its attention. In parallel, 'Feature visualization' offers another avenue for interpreting neural network features. This approach synthesizes an optimal image through gradient ascent, providing clearer insights into 'what' features respond to. However, feature visualizations only provide one global explanation per feature; they do not explain why features activate for particular images. In this work, we introduce a new method to the interpretability tool-kit, 'feature accentuation', which is capable of conveying both 'where' and 'what' in arbitrary input images induces a feature's response. At its core, feature accentuation is image-seeded (rather than noise-seeded) feature visualization. We find a particular combination of parameterization, augmentation, and regularization yields naturalistic visualizations that resemble the seed image and target feature simultaneously. Furthermore, we validate these accentuations are processed along a natural circuit by the model. We make our precise implementation of 'feature accentuation' available to the community as the 'Faccent' library, an extension of the popular 'Lucent' library for feature visualization.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 2666