On convex decision regions in deep network representations

• URL: http://arxiv.org/abs/2305.17154v2
• Date: Fri, 6 Oct 2023 14:58:58 GMT
• Title: On convex decision regions in deep network representations
• Authors: Lenka T\v{e}tkov\'a, Thea Br\"usch, Teresa Karen Scheidt, Fabian Martin Mager, Rasmus {\O}rtoft Aagaard, Jonathan Foldager, Tommy Sonne Alstr{\o}m and Lars Kai Hansen
• Abstract summary: We investigate the notion of convexity of concept regions in machine-learned latent spaces. We show that convexity is robust to basic re-parametrization. We find that approximate convexity is pervasive in neural representations in multiple application domains.
• Score: 1.06378109904813
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Current work on human-machine alignment aims at understanding machine-learned latent spaces and their correspondence to human representations. G{\"a}rdenfors' conceptual spaces is a prominent framework for understanding human representations. Convexity of object regions in conceptual spaces is argued to promote generalizability, few-shot learning, and interpersonal alignment. Based on these insights, we investigate the notion of convexity of concept regions in machine-learned latent spaces. We develop a set of tools for measuring convexity in sampled data and evaluate emergent convexity in layered representations of state-of-the-art deep networks. We show that convexity is robust to basic re-parametrization and, hence, meaningful as a quality of machine-learned latent spaces. We find that approximate convexity is pervasive in neural representations in multiple application domains, including models of images, audio, human activity, text, and medical images. Generally, we observe that fine-tuning increases the convexity of label regions. We find evidence that pretraining convexity of class label regions predicts subsequent fine-tuning performance.

Related papers

• Space Explanations of Neural Network Classification [2.823533769284529]
  We present a novel logic-based concept called Space Explanations for classifying neural networks.<n>To automatically generate space explanations, we leverage a range of flexible Craig algorithms and unsatisfiable core generation.
  arXiv  Detail & Related papers  (2025-11-27T14:33:59Z)
• Into the Rabbit Hull: From Task-Relevant Concepts in DINO to Minkowski Geometry [31.26429968473424]
  DINOv2 is routinely deployed to recognize objects, scenes, and actions; yet the nature of what it perceives remains unknown.<n>As a working baseline, we adopt the Linear Representation Hypothesis (LRH) and operationalize it using SAEs.<n>We produce a 32,000-unit dictionary that serves as the interpretability backbone of our study.
  arXiv  Detail & Related papers  (2025-10-08T22:42:20Z)
• Generalized Decoupled Learning for Enhancing Open-Vocabulary Dense Perception [71.26728044621458]
  DeCLIP is a novel framework that enhances CLIP by decoupling the self-attention module to obtain content'' and context'' features respectively.<n>It consistently achieves state-of-the-art performance across a broad spectrum of tasks, including 2D detection and segmentation, 3D instance segmentation, video instance segmentation, and 6D object pose estimation.
  arXiv  Detail & Related papers  (2025-08-15T06:43:51Z)
• Concept Probing: Where to Find Human-Defined Concepts (Extended Version) [3.2443914909457594]
  We propose a method to automatically identify which layer's representations in a neural network model should be considered when probing for a given human-defined concept of interest.<n>We validate our findings through an exhaustive empirical analysis over different neural network models and datasets.
  arXiv  Detail & Related papers  (2025-07-24T16:30:10Z)
• FeatInv: Spatially resolved mapping from feature space to input space using conditional diffusion models [0.9503773054285559]
  Internal representations are crucial for understanding deep neural networks.<n>While mapping from feature space to input space aids in interpreting the former, existing approaches often rely on crude approximations.<n>We propose using a conditional diffusion model to learn such a mapping in a probabilistic manner.
  arXiv  Detail & Related papers  (2025-05-27T11:07:34Z)
• The Origins of Representation Manifolds in Large Language Models [52.68554895844062]
  We show that cosine similarity in representation space may encode the intrinsic geometry of a feature through shortest, on-manifold paths.<n>The critical assumptions and predictions of the theory are validated on text embeddings and token activations of large language models.
  arXiv  Detail & Related papers  (2025-05-23T13:31:22Z)
• Exploring Geometric Representational Alignment through Ollivier-Ricci Curvature and Ricci Flow [0.0]
  We use Ollivier-Ricci curvature and Ricci flow as tools to study the alignment of representations between humans and artificial neural systems. As a proof-of-principle study, we compared the representations of face stimuli between VGG-Face, a human-aligned version of VGG-Face, and corresponding human similarity judgments from a large online study.
  arXiv  Detail & Related papers  (2025-01-01T18:33:48Z)
• Connecting Concept Convexity and Human-Machine Alignment in Deep Neural Networks [3.001674556825579]
  Understanding how neural networks align with human cognitive processes is a crucial step toward developing more interpretable and reliable AI systems. We identify a correlation between these two dimensions that reflect the similarity relations humans in cognitive tasks. This presents a first step toward understanding the relationship convexity between human-machine alignment.
  arXiv  Detail & Related papers  (2024-09-10T09:32:16Z)
• Understanding Distributed Representations of Concepts in Deep Neural Networks without Supervision [25.449397570387802]
  We propose an unsupervised method for discovering distributed representations of concepts by selecting a principal subset of neurons. Our empirical findings demonstrate that instances with similar neuron activation states tend to share coherent concepts. It can be utilized to identify unlabeled subclasses within data and to detect the causes of misclassifications.
  arXiv  Detail & Related papers  (2023-12-28T07:33:51Z)
• Rewrite Caption Semantics: Bridging Semantic Gaps for Language-Supervised Semantic Segmentation [100.81837601210597]
  We propose Concept Curation (CoCu) to bridge the gap between visual and textual semantics in pre-training data. CoCu achieves superb zero-shot transfer performance and greatly boosts language-supervised segmentation baseline by a large margin.
  arXiv  Detail & Related papers  (2023-09-24T00:05:39Z)
• SimNP: Learning Self-Similarity Priors Between Neural Points [52.4201466988562]
  SimNP is a method to learn category-level self-similarities. We show that SimNP is able to outperform previous methods in reconstructing symmetric unseen object regions.
  arXiv  Detail & Related papers  (2023-09-07T16:02:40Z)
• Denoise and Contrast for Category Agnostic Shape Completion [48.66519783934386]
  We present a deep learning model that exploits the power of self-supervision to perform 3D point cloud completion. A denoising pretext task provides the network with the needed local cues, decoupled from the high-level semantics. contrastive learning maximizes the agreement between variants of the same shape with different missing portions.
  arXiv  Detail & Related papers  (2021-03-30T20:33:24Z)
• Variational Structured Attention Networks for Deep Visual Representation Learning [49.80498066480928]
  We propose a unified deep framework to jointly learn both spatial attention maps and channel attention in a principled manner. Specifically, we integrate the estimation and the interaction of the attentions within a probabilistic representation learning framework. We implement the inference rules within the neural network, thus allowing for end-to-end learning of the probabilistic and the CNN front-end parameters.
  arXiv  Detail & Related papers  (2021-03-05T07:37:24Z)
• Regional Attention Network (RAN) for Head Pose and Fine-grained Gesture Recognition [9.131161856493486]
  We propose a novel end-to-end textbfRegional Attention Network (RAN), which is a fully Convolutional Neural Network (CNN) Our regions consist of one or more consecutive cells and are adapted from the strategies used in computing HOG (Histogram of Oriented Gradient) descriptor. The proposed approach outperforms the state-of-the-art by a considerable margin in different metrics.
  arXiv  Detail & Related papers  (2021-01-17T10:14:28Z)
• Evidential Sparsification of Multimodal Latent Spaces in Conditional Variational Autoencoders [63.46738617561255]
  We consider the problem of sparsifying the discrete latent space of a trained conditional variational autoencoder. We use evidential theory to identify the latent classes that receive direct evidence from a particular input condition and filter out those that do not. Experiments on diverse tasks, such as image generation and human behavior prediction, demonstrate the effectiveness of our proposed technique.
  arXiv  Detail & Related papers  (2020-10-19T01:27:21Z)
• 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)
• Focus on Semantic Consistency for Cross-domain Crowd Understanding [34.560447389853614]
  Some domain adaptation algorithms try to liberate it by training models with synthetic data. We found that a mass of estimation errors in the background areas impede the performance of the existing methods. In this paper, we propose a domain adaptation method to eliminate it.
  arXiv  Detail & Related papers  (2020-02-20T08:51:05Z)
• Topologically Densified Distributions [25.140319008330167]
  We study regularization in the context of small sample-size learning with over- parameterized neural networks. We impose a topological constraint on samples drawn from the probability measure induced in that space. This provably leads to mass concentration effects around the representations of training instances.
  arXiv  Detail & Related papers  (2020-02-12T05:25:15Z)

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.