Generative Adversarial Networks and Image Synthesis

Derniers papiers

• Toward Identifiable Sparse Autoencoders

  Walter Nelson, Theofanis Karaletsos, Francesco Locatello · 1 juin 2026

  Recently, sparse autoencoders (SAEs) have emerged as an attractive tool for interpreting and interacting with representations in practical neural networks. While it is common empirical folklore, we also show theoretically that SAEs are highly unstable: different training runs are likely to produce d…

• On the regularization of Wasserstein GANs

  Henning Petzka, Asja Fischer, Denis Lukovnikov · 1 juin 2026

  Since their invention, generative adversarial networks (GANs) have become a popular approach for learning to model a distribution of real (unlabeled) data. Convergence problems during training are overcome by Wasserstein GANs which minimize the distance between the model and the empirical distributi…

• Improving Relative Representations with Learned Anchors and Whitened Inner Products

  Oscar Thorsted Svendsen, Nikolaj Holst Jakobsen, Fabian Mager, Hiba Nassar · 1 juin 2026

  Independently trained neural models typically converge to incompatible latent representations, creating a fundamental barrier to highly modular AI systems. While Relative Representations (RR) address this by mapping absolute coordinates to a shared space defined by similarities to common anchor poin…

• Riemannian Diffusion Models on General Manifolds via Physics-Informed Neural Networks

  Gyeonghoon Ko, Juho Lee · 1 juin 2026

  Riemannian diffusion models generalize score-based generative modeling to manifold-supported data via stochastic diffusion equations on the manifold. However, training requires sampling from and differentiating the manifold heat kernel, which is rarely available in closed form beyond a few highly sy…

• Equivariant Latent Alignment via Flow Matching under Group Symmetries

  Sunghyun Kim, Jaehoon Hahm, Jeongwoo Shin, Joonseok Lee · 1 juin 2026

  Geometry-aware generative models and novel view synthesis approaches have shown strong potential in visual fidelity and consistency. In parallel, equivariant representation learning has emerged as a powerful framework for constructing latent spaces where analytically known group transformations coul…

• Memory by Design: Probabilistic Sequence Layers

  Matthew Dowling, Hyungju Jeon, Cristina Savin, Il Memming Park · 1 juin 2026

  We introduce the design-model framework: a way to derive efficient recurrent sequence maps from explicit assumptions about memory. A design model writes evidence into memory by exact Bayesian filtering; a query-dependent readout produces a predictive distribution whose mean is the layer output. In o…

• Envisioning Beyond the Few: Disentangled Semantics and Primitives for Few-Shot Atypical Layout-to-Image Generation

  Nan Bao, Yifan Zhao, Wenzhuang Wang, Jia Li · 1 juin 2026

  The layout-to-image (L2I) task enables fine-grained control over image generation via object categories and spatial layouts. However, existing L2I methods yield fragmented and distorted generations under few-shot atypical settings. We term this failure as representation fragmentation, arising from a…

• A Kinetic Energy Perspective of Flow Matching

  Ziyun Li, Huancheng Hu, Soon Hoe Lim, Xuyu Li, Fei Gao, Enmao Diao, Zezhen Ding, Michalis Vazirgiannis, Henrik Bostrom · 1 juin 2026

  Flow-based generative models can be viewed through a physics lens: sampling transports a particle from noise to data by integrating a learned velocity field, and each sample corresponds to a trajectory with its own dynamical effort. Motivated by classical mechanics, we introduce Kinetic Path Energy …

• Diffusion Models Preferentially Memorize Prototypical Examples or: Why Does My Diffusion Model Love Slop?

  Marta Aparicio Rodriguez, Anastasia Borovykh, Grigorios A. Pavliotis, Daniel J. Korchinski · 1 juin 2026

  Generative models have a persistent limitation: their tendency to memorize training data can create legal liabilities and erode creative diversity. Understanding which samples are memorized in whole or in part, and under what conditions, therefore remains an important open problem. Here we answer th…

• Augmented Lagrangian Predictive Coding

  Jeffrey Seely, Julian Gould · 1 juin 2026

  Predictive coding (PC) is a local-learning alternative to backpropagation (BP), training deep networks via local energy-minimization dynamics rather than a global backward pass. We introduce Augmented Lagrangian Predictive Coding (PC-ALM), which maintains PC's inference budget but aligns each weight…

• Scalable Inference-Time Annealing with Surrogate Likelihood Estimators

  Daniel Pe\~naherrera, Rishal Aggarwal, David Ryan Koes · 1 juin 2026

  A long standing challenge in computational chemistry and biophysics is efficiently sampling the Boltzmann distribution of molecules. Advances in generative modeling have been proposed to address the limitations of conventional sampling techniques by eliminating the computational cost of simulation. …

• TunerDiT: Training-free Progressive Steering of Diffusion Transformer for Multi-Event Video Generation

  Ruotong Liao, Guowen Huang, Qing Cheng, Guangyao Zhai, Lei Zhang, Xun Xiao, Thomas Seidl, Daniel Cremers, Volker Tresp · 1 juin 2026

  Text-to-video (T2V) generation faces challenging questions when generating videos with long horizons containing multiple events. Inspired by the intrinsics of the diffusion process, we probe video diffusion transformers (DiTs) and uncover intrinsic turning points in the DiT denoising trajectory wher…

• Softly Constrained Denoisers for Diffusion Models Applied to Partial Differential Equations

  Victor M. Yeom-Song, Severi Rissanen, Arno Solin, Samuel Kaski, Mingfei Sun · 1 juin 2026

  Diffusion models have become a powerful generative prior for solutions of partial differential equations (PDEs). Existing approaches enforce physical constraints either by adding the PDE residuals as loss regularizers or through inference-time adjustments. These methods bias the model away from the …

• LVSA: Training-Free Sparse Attention for Long Video Diffusion

  Gael Glorian, Ioannis Lamprou, Zhen Zhang, Yujie Yuan, Hongsheng Liu · 1 juin 2026

  Dense self-attention is the compute and quality bottleneck of long-video diffusion inference: cost grows quadratically with the sequence length, and beyond the training horizon the model converges to near-static output, that is, "frozen" repetitive video. State of the art approaches are either too c…

• RayDer: Scalable Self-Supervised Novel View Synthesis from Real-World Video

  Ulrich Prestel, Stefan Andreas Baumann, Nick Stracke, Bj\"orn Ommer · 1 juin 2026

  Self-supervised novel view synthesis (NVS) remains challenging to scale, despite the abundance of video data, largely due to the brittleness of training on realistic videos and the hard-to-predict scaling behavior of multi-network system designs. We introduce RayDer, a unified, feed-forward transfor…

• Vision-Language Models Suppress Female Representations Under Ambiguous Input

  Arnau Marin-Llobet, Simon Henniger, Mahzarin R. Banaji · 1 juin 2026

  Alignment teaches vision-language models (VLMs) to avoid expressing demographic biases, and when gender is clearly visible they largely succeed. Far less is known about ambiguous inputs (a worker in full gear, a figure seen from behind) cases common in practice yet rarely studied. We find that minim…

• On the Relationship Between Activation Outliers and Feature Death in Sparse Autoencoders

  Elana Simon, Etowah Adams, James Zou · 1 juin 2026

  Sparse autoencoders (SAEs) decompose neural network activations into interpretable features, but many learned features never activate, a problem called feature death that wastes dictionary capacity and can reintroduce superposition. Death rates vary dramatically between models: near-zero on GPT-2, o…

• Lumos-Nexus: Efficient Frequency Bridging with Homogeneous Latent Space for Video Unified Models

  Jiazheng Xing, Hangjie Yuan, Lingling Cai, Xinyu Liu, Yujie Wei, Fei Du, Hai Ci, Tao Feng, Jiasheng Tang, Weihua Chen, Fan Wang, Yong Liu · 1 juin 2026

  Connector-based video unified models have demonstrated strong capability in instruction-grounded video synthesis, but integrating a large high-fidelity generator into the unified training loop is computationally prohibitive, limiting achievable visual quality. We therefore propose Lumos-Nexus, a tra…

• Conformal Reliability: A New Evaluation Metric for Conditional Generation

  Yachen Gao, Xinwei Sun, Yikai Wang, Ye Shi, Jingya Wang, Jianfeng Feng, Yanwei Fu · 1 juin 2026

  Conditional generative models have recently achieved remarkable success in various applications. However, a suitable metric for evaluating the reliability of these models, which takes into account their inherent uncertainty, is still lacking. Existing metrics, which typically assess a single output,…

• A Unifying View of Variational Generative Wasserstein Flows

  Paul Caucheteux, Cl\'ement Bonet, Anna Korba · 1 juin 2026

  Many modern generative models can be viewed as minimizing divergences between probability distributions, yet they rely on different algorithmic and geometric principles. Wasserstein gradient flows provide a continuous-time formulation for optimizing over distributions, and can be approximated throug…

• Parallel Tempering Initial Sampling in Inference-Time Reward Alignment

  Myeongjun Oh, Gwangho Kim, Sungyoon Lee · 1 juin 2026

  Inference-time reward alignment steers pretrained diffusion and flow-based generative models to satisfy user-specified rewards without retraining. Recently, Sequential Monte Carlo (SMC) has emerged as a powerful framework for this task by iteratively filtering and propagating multiple particles. How…

• Fixed-Point Masked Generative Modeling

  Andrea Miele, Yiming Qin, Alba Carballo-Castro, Justin Deschenaux, Pascal Frossard · 1 juin 2026

  Masked Generative Models (MGMs) enable parallel decoding and achieve strong performance across modalities, but require full-sequence bidirectional transformers at every step, making training costly and degrading quality under low sampling budgets. Existing work improves efficiency via better sampler…

• Generative Models and Statistical Validation

  Sascha Diefenbacher, Sofia Palacios Schweitzer, Gregor Kasieczka · 1 juin 2026

  Generative machine learning has become an essential tool in theoretical and experimental physics, especially in the context of fast surrogates and density estimators. In this work, we first introduce the underlying framework of modern generative networks and then discuss challenges in quantifying th…

• Unlearning in Diffusion Models: A Unified Framework with KL Divergence and Likelihood Constraints

  Shervin Khalafi, Alejandro Ribeiro, Dongsheng Ding · 1 juin 2026

  Unlearning in diffusion models aims to remove undesirable data or concepts while preserving the utility of pretrained models -- two fundamentally conflicting objectives. We propose a principled constrained optimization framework that formulates unlearning as minimizing the deviation from a pretraine…

• Gradient Preconditioning for Efficient and Reliable Reward-Guided Generation

  Jisung Hwang, Minhyuk Sung · 29 mai 2026

  We propose a gradient preconditioning method that makes reward-guided generation with one-step generative models both efficient and reliable. Test-time noise optimization can unlock substantially better reward-guided generations from pretrained generative models, but it is prone to reward hacking th…