Feature Accentuation: Revealing 'What' Features Respond to in Natural Images

• URL: http://arxiv.org/abs/2402.10039v2
• Date: Sun, 9 Jun 2024 03:06:59 GMT
• Title: Feature Accentuation: Revealing 'What' Features Respond to in Natural Images
• Authors: Chris Hamblin, Thomas Fel, Srijani Saha, Talia Konkle, George Alvarez,
• Abstract summary: 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. We find a particular combination of parameterization, augmentation, and regularization yields naturalistic visualizations that resemble the seed image and target feature simultaneously. We make our precise implementation of feature accentuation available to the community as the Faccent library, an extension of Lucent.
• Score: 4.4273123155989715
• License: http://creativecommons.org/licenses/by/4.0/
• 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 Lucent.

Related papers

• Cross-Image Attention for Zero-Shot Appearance Transfer [68.43651329067393]
  We introduce a cross-image attention mechanism that implicitly establishes semantic correspondences across images. We harness three mechanisms that either manipulate the noisy latent codes or the model's internal representations throughout the denoising process. Experiments show that our method is effective across a wide range of object categories and is robust to variations in shape, size, and viewpoint.
  arXiv  Detail & Related papers  (2023-11-06T18:33:24Z)
• Unlocking Feature Visualization for Deeper Networks with MAgnitude Constrained Optimization [17.93878159391899]
  We describe MACO, a simple approach to generate interpretable images. Our approach yields significantly better results (both qualitatively and quantitatively) and unlocks efficient and interpretable feature visualizations for large state-of-the-art neural networks. We validate our method on a novel benchmark for comparing feature visualization methods, and release its visualizations for all classes of the ImageNet dataset.
  arXiv  Detail & Related papers  (2023-06-11T23:33:59Z)
• Don't trust your eyes: on the (un)reliability of feature visualizations [25.018840023636546]
  We show how to trick feature visualizations into showing arbitrary patterns that are completely disconnected from normal network behavior on natural input. We then provide evidence for a similar phenomenon occurring in standard, unmanipulated networks. This can be used as a sanity check for feature visualizations.
  arXiv  Detail & Related papers  (2023-06-07T18:31:39Z)
• GLANCE: Global to Local Architecture-Neutral Concept-based Explanations [26.76139301708958]
  We propose a novel twin-surrogate explainability framework to explain the decisions made by any CNN-based image classifier. We first disentangle latent features from the classifier, followed by aligning these features to observed/human-defined context' features. These aligned features form semantically meaningful concepts that are used for extracting a causal graph depicting the perceived' data-generating process. We provide a generator to visualize the effect' of interactions among features in latent space and draw feature importance therefrom as local explanations.
  arXiv  Detail & Related papers  (2022-07-05T09:52:09Z)
• Attribute Prototype Network for Any-Shot Learning [113.50220968583353]
  We argue that an image representation with integrated attribute localization ability would be beneficial for any-shot, i.e. zero-shot and few-shot, image classification tasks. We propose a novel representation learning framework that jointly learns global and local features using only class-level attributes.
  arXiv  Detail & Related papers  (2022-04-04T02:25:40Z)
• Semantic Disentangling Generalized Zero-Shot Learning [50.259058462272435]
  Generalized Zero-Shot Learning (GZSL) aims to recognize images from both seen and unseen categories. In this paper, we propose a novel feature disentangling approach based on an encoder-decoder architecture. The proposed model aims to distill quality semantic-consistent representations that capture intrinsic features of seen images.
  arXiv  Detail & Related papers  (2021-01-20T05:46:21Z)
• Attribute Prototype Network for Zero-Shot Learning [113.50220968583353]
  We propose a novel zero-shot representation learning framework that jointly learns discriminative global and local features. Our model points to the visual evidence of the attributes in an image, confirming the improved attribute localization ability of our image representation.
  arXiv  Detail & Related papers  (2020-08-19T06:46:35Z)
• Saliency-driven Class Impressions for Feature Visualization of Deep Neural Networks [55.11806035788036]
  It is advantageous to visualize the features considered to be essential for classification. Existing visualization methods develop high confidence images consisting of both background and foreground features. In this work, we propose a saliency-driven approach to visualize discriminative features that are considered most important for a given task.
  arXiv  Detail & Related papers  (2020-07-31T06:11:06Z)
• Geometrically Mappable Image Features [85.81073893916414]
  Vision-based localization of an agent in a map is an important problem in robotics and computer vision. We propose a method that learns image features targeted for image-retrieval-based localization.
  arXiv  Detail & Related papers  (2020-03-21T15:36:38Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.