Generate in Reconstruction Space, Match in Semantic Space: Transport Geometry for One-Step Generation

• URL: http://arxiv.org/abs/2606.00514v1
• Date: Sat, 30 May 2026 04:03:47 GMT
• Title: Generate in Reconstruction Space, Match in Semantic Space: Transport Geometry for One-Step Generation
• Abstract summary: Generative modeling and self-supervised representation learning (SSL) optimize structurally different objectives.<n>We study the benefits of SSL in generative modeling in the framework of one-step generation.<n>We find that this objective becomes highly effective when computed in a semantically structured SSL feature space.
• Score: 16.138914735031744
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Generative modeling and self-supervised representation learning (SSL) optimize structurally different objectives: generative training rewards distributional fidelity, while SSL rewards semantic coherence. Yet recent work repeatedly finds that SSL features improve generative training, though the mechanism of this synergy remains unclear. Here, we study the benefits of SSL in generative modeling in the framework of one-step generation where the role of representation is explicit: frozen SSL features are used to match generated samples to real data. We use the Sinkhorn divergence in that feature space, providing a tractable surrogate for the Wasserstein distance, the population-level discrepancy approximated by Fréchet-style evaluation metrics (such as FID). We find that this objective becomes highly effective when computed in a semantically structured SSL feature space (a 39$\times$ reduction in ImageNet FID). We trace this behavior primarily to matching estimation: semantic SSL features that suppress nuisance reconstruction details induce a more compact geometry, making distribution matching more tractable. As a consequence, the best training SSL features need not match the features used by the evaluation metric. In particular, we show that using Inception as the feature extractor can improve FID while degrading matching stability and sample quality, revealing a form of metric hacking. Using extensive experiments on ImageNet, we identify which SSL feature families lead to best generation performance and show that matching stability is a quantitative criterion for selecting them. Code is available at https://github.com/Genentech/semantic-transport-generation.

Related papers

• Self-Supervised Representation-Guided Generative Dataset Distillation [55.27114962330541]
  We propose self-supervised representation-guided generative dataset distillation (SRG)<n>SRG translates the SSL geometry into diffusion guidance.<n>It consistently outperforms evaluated generative baselines across multiple datasets and IPC settings.
  arXiv  Detail & Related papers  (2026-08-04T06:51:45Z)
• Probing Geospatial SSL Representations with Environmental Signals [51.908606610432194]
  Self-supervised learning (SSL) is designed to learn generic, transferable representations rather than representations optimized for a single task.<n>We ask a different question: do SSL representations preserve statistical associations with environmental variables that co-vary with the imaging process?<n>To answer this question, we probe SSL representations using co-located ERA5 reanalysis variables, a global dataset of physically consistent environmental variables.
  arXiv  Detail & Related papers  (2026-07-06T15:24:25Z)
• Learning Sparse Visual Representations via Spatial-Semantic Factorization [37.169502692169196]
  Self-supervised learning (SSL) faces a fundamental conflict between semantic understanding and image reconstruction.<n>We introduce STELLAR, a framework that factorizes visual features into a low-rank product of semantic concepts and their spatial distributions.<n>We demonstrate that as few as 16 sparse tokens under this factorized form are sufficient to simultaneously support high-quality reconstruction (2.60 FID) and match the semantic performance of dense backbones (79.10% ImageNet accuracy)
  arXiv  Detail & Related papers  (2026-02-02T10:12:17Z)
• Semantic Concentration for Self-Supervised Dense Representations Learning [103.10708947415092]
  Image-level self-supervised learning (SSL) has made significant progress, yet learning dense representations for patches remains challenging.<n>This work reveals that image-level SSL avoids over-dispersion by involving implicit semantic concentration.
  arXiv  Detail & Related papers  (2025-09-11T13:12:10Z)
• On the Discriminability of Self-Supervised Representation Learning [38.598160031349686]
  Self-supervised learning (SSL) has recently shown notable success in various visual tasks.<n>However, in terms of discriminability, SSL is still not on par with supervised learning (SL)<n>This paper identifies a key issue, the crowding problem," where features from different classes are not well-separated.
  arXiv  Detail & Related papers  (2024-07-18T14:18:03Z)
• Benchmark for Uncertainty & Robustness in Self-Supervised Learning [0.0]
  Self-Supervised Learning is crucial for real-world applications, especially in data-hungry domains such as healthcare and self-driving cars. In this paper, we explore variants of SSL methods, including Jigsaw Puzzles, Context, Rotation, Geometric Transformations Prediction for vision, as well as BERT and GPT for language tasks. Our goal is to create a benchmark with outputs from experiments, providing a starting point for new SSL methods in Reliable Machine Learning.
  arXiv  Detail & Related papers  (2022-12-23T15:46:23Z)
• Improving Self-Supervised Learning by Characterizing Idealized Representations [155.1457170539049]
  We prove necessary and sufficient conditions for any task invariant to given data augmentations. For contrastive learning, our framework prescribes simple but significant improvements to previous methods. For non-contrastive learning, we use our framework to derive a simple and novel objective.
  arXiv  Detail & Related papers  (2022-09-13T18:01:03Z)
• GSMFlow: Generation Shifts Mitigating Flow for Generalized Zero-Shot Learning [55.79997930181418]
  Generalized Zero-Shot Learning aims to recognize images from both the seen and unseen classes by transferring semantic knowledge from seen to unseen classes. It is a promising solution to take the advantage of generative models to hallucinate realistic unseen samples based on the knowledge learned from the seen classes. We propose a novel flow-based generative framework that consists of multiple conditional affine coupling layers for learning unseen data generation.
  arXiv  Detail & Related papers  (2022-07-05T04:04:37Z)
• InsCLR: Improving Instance Retrieval with Self-Supervision [30.36455490844235]
  We find that fine-tuning using the recently developed self-supervised (SSL) learning methods, such as SimCLR and MoCo, fails to improve the performance of instance retrieval. To overcome this problem, we propose InsCLR, a new SSL method that builds on the textitinstance-level contrast. InsCLR achieves similar or even better performance than the state-of-the-art SSL methods on instance retrieval.
  arXiv  Detail & Related papers  (2021-12-02T16:21:27Z)
• Self-Supervised Learning of Graph Neural Networks: A Unified Review [50.71341657322391]
  Self-supervised learning is emerging as a new paradigm for making use of large amounts of unlabeled samples. We provide a unified review of different ways of training graph neural networks (GNNs) using SSL. Our treatment of SSL methods for GNNs sheds light on the similarities and differences of various methods, setting the stage for developing new methods and algorithms.
  arXiv  Detail & Related papers  (2021-02-22T03:43:45Z)
• Interventional Few-Shot Learning [88.31112565383457]
  We propose a novel Few-Shot Learning paradigm: Interventional Few-Shot Learning. Code is released at https://github.com/yue-zhongqi/ifsl.
  arXiv  Detail & Related papers  (2020-09-28T01:16:54Z)
• Information Bottleneck Constrained Latent Bidirectional Embedding for Zero-Shot Learning [59.58381904522967]
  We propose a novel embedding based generative model with a tight visual-semantic coupling constraint. We learn a unified latent space that calibrates the embedded parametric distributions of both visual and semantic spaces. Our method can be easily extended to transductive ZSL setting by generating labels for unseen images.
  arXiv  Detail & Related papers  (2020-09-16T03:54:12Z)
• Semi-supervised learning objectives as log-likelihoods in a generative model of data curation [32.45282187405337]
  We formulate SSL objectives as a log-likelihood in a generative model of data curation. We give a proof-of-principle for Bayesian SSL on toy data.
  arXiv  Detail & Related papers  (2020-08-13T13:50:27Z)

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.