Beyond Scalars: Concept-Based Alignment Analysis in Vision Transformers

• URL: http://arxiv.org/abs/2412.06639v1
• Date: Mon, 09 Dec 2024 16:33:28 GMT
• Title: Beyond Scalars: Concept-Based Alignment Analysis in Vision Transformers
• Authors: Johanna Vielhaben, Dilyara Bareeva, Jim Berend, Wojciech Samek, Nils Strodthoff,
• Abstract summary: Vision transformers (ViTs) can be trained using various learning paradigms, from fully supervised to self-supervised. We propose a concept-based alignment analysis of representations from four different ViTs. The concept-based alignment analysis of representations from four different ViTs reveals that increased supervision correlates with a reduction in the semantic structure of learned representations.
• Score: 10.400355814467401
• License:
• Abstract: Vision transformers (ViTs) can be trained using various learning paradigms, from fully supervised to self-supervised. Diverse training protocols often result in significantly different feature spaces, which are usually compared through alignment analysis. However, current alignment measures quantify this relationship in terms of a single scalar value, obscuring the distinctions between common and unique features in pairs of representations that share the same scalar alignment. We address this limitation by combining alignment analysis with concept discovery, which enables a breakdown of alignment into single concepts encoded in feature space. This fine-grained comparison reveals both universal and unique concepts across different representations, as well as the internal structure of concepts within each of them. Our methodological contributions address two key prerequisites for concept-based alignment: 1) For a description of the representation in terms of concepts that faithfully capture the geometry of the feature space, we define concepts as the most general structure they can possibly form - arbitrary manifolds, allowing hidden features to be described by their proximity to these manifolds. 2) To measure distances between concept proximity scores of two representations, we use a generalized Rand index and partition it for alignment between pairs of concepts. We confirm the superiority of our novel concept definition for alignment analysis over existing linear baselines in a sanity check. The concept-based alignment analysis of representations from four different ViTs reveals that increased supervision correlates with a reduction in the semantic structure of learned representations.

Related papers

• Towards a Learning Theory of Representation Alignment [12.166663160280056]
  We propose a learning-theoretic perspective to representation alignment. Our results can be seen as a first step toward casting representation alignment as a learning-theoretic problem.
  arXiv  Detail & Related papers  (2025-02-19T19:09:14Z)
• Local Concept Embeddings for Analysis of Concept Distributions in DNN Feature Spaces [1.0923877073891446]
  We propose a novel concept analysis framework for deep neural networks (DNNs) Instead of optimizing a single global concept vector on the complete dataset, it generates a local concept embedding (LoCE) vector for each individual sample. Despite its context sensitivity, our method's concept segmentation performance is competitive to global baselines.
  arXiv  Detail & Related papers  (2023-11-24T12:22:00Z)
• Identifying Linear Relational Concepts in Large Language Models [16.917379272022064]
  Transformer language models (LMs) have been shown to represent concepts as directions in the latent space of hidden activations. We present a technique called linear relational concepts (LRC) for finding concept directions corresponding to human-interpretable concepts.
  arXiv  Detail & Related papers  (2023-11-15T14:01:41Z)
• Disentangling Multi-view Representations Beyond Inductive Bias [32.15900989696017]
  We propose a novel multi-view representation disentangling method that ensures both interpretability and generalizability of the resulting representations. Our experiments on four multi-view datasets demonstrate that our proposed method outperforms 12 comparison methods in terms of clustering and classification performance.
  arXiv  Detail & Related papers  (2023-08-03T09:09:28Z)
• Identifying Interpretable Subspaces in Image Representations [54.821222487956355]
  We propose a framework to explain features of image representations using Contrasting Concepts (FALCON) For a target feature, FALCON captions its highly activating cropped images using a large captioning dataset and a pre-trained vision-language model like CLIP. Each word among the captions is scored and ranked leading to a small number of shared, human-understandable concepts.
  arXiv  Detail & Related papers  (2023-07-20T00:02:24Z)
• A Probabilistic Relaxation of the Two-Stage Object Pose Estimation Paradigm [0.0]
  We propose a matching-free probabilistic formulation for object pose estimation. It enables unified and concurrent optimization of both visual correspondence and geometric alignment. It can represent different plausible modes of the entire distribution of likely poses.
  arXiv  Detail & Related papers  (2023-06-01T16:50:40Z)
• Progressive Semantic-Visual Mutual Adaption for Generalized Zero-Shot Learning [74.48337375174297]
  Generalized Zero-Shot Learning (GZSL) identifies unseen categories by knowledge transferred from the seen domain. We deploy the dual semantic-visual transformer module (DSVTM) to progressively model the correspondences between prototypes and visual features. DSVTM devises an instance-motivated semantic encoder that learns instance-centric prototypes to adapt to different images, enabling the recast of the unmatched semantic-visual pair into the matched one.
  arXiv  Detail & Related papers  (2023-03-27T15:21:43Z)
• Measuring the Interpretability of Unsupervised Representations via Quantized Reverse Probing [97.70862116338554]
  We investigate the problem of measuring interpretability of self-supervised representations. We formulate the latter as estimating the mutual information between the representation and a space of manually labelled concepts. We use our method to evaluate a large number of self-supervised representations, ranking them by interpretability.
  arXiv  Detail & Related papers  (2022-09-07T16:18:50Z)
• Concept Gradient: Concept-based Interpretation Without Linear Assumption [77.96338722483226]
  Concept Activation Vector (CAV) relies on learning a linear relation between some latent representation of a given model and concepts. We proposed Concept Gradient (CG), extending concept-based interpretation beyond linear concept functions. We demonstrated CG outperforms CAV in both toy examples and real world datasets.
  arXiv  Detail & Related papers  (2022-08-31T17:06:46Z)
• Toward a Visual Concept Vocabulary for GAN Latent Space [74.12447538049537]
  This paper introduces a new method for building open-ended vocabularies of primitive visual concepts represented in a GAN's latent space. Our approach is built from three components: automatic identification of perceptually salient directions based on their layer selectivity; human annotation of these directions with free-form, compositional natural language descriptions. Experiments show that concepts learned with our approach are reliable and composable -- generalizing across classes, contexts, and observers.
  arXiv  Detail & Related papers  (2021-10-08T17:58:19Z)
• Closed-Form Factorization of Latent Semantics in GANs [65.42778970898534]
  A rich set of interpretable dimensions has been shown to emerge in the latent space of the Generative Adversarial Networks (GANs) trained for synthesizing images. In this work, we examine the internal representation learned by GANs to reveal the underlying variation factors in an unsupervised manner. We propose a closed-form factorization algorithm for latent semantic discovery by directly decomposing the pre-trained weights.
  arXiv  Detail & Related papers  (2020-07-13T18:05:36Z)

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.