Generative Modeling by Estimating Gradients of the Data Distribution
2019/07/12 by Yang Song, Stefano Ermon, Song, Yang +1 · 1 voice · 461 citations
Biochemistry, Genetics and Molecular Biology · Computer Science · Mathematics · #Anomaly Detection Techniques and Applications #Cell Image Analysis Techniques #Generative Adversarial Networks and Image Synthesis #cs.LG #stat.ML
paper · pdf · doi:10.48550/arxiv.1907.05600
arxiv published 2019/07/12 · arxiv updated 2020/10/10
Abstract
We introduce a new generative model where samples are produced via Langevin dynamics using gradients of the data distribution estimated with score matching. Because gradients can be ill-defined and hard to estimate when the data resides on low-dimensional manifolds, we perturb the data with different levels of Gaussian noise, and jointly estimate the corresponding scores, i.e., the vector fields of gradients of the perturbed data distribution for all noise levels. For sampling, we propose an annealed Langevin dynamics where we use gradients corresponding to gradually decreasing noise levels as the sampling process gets closer to the data manifold. Our framework allows flexible model architectures, requires no sampling during training or the use of adversarial methods, and provides a learning objective that can be used for principled model comparisons. Our models produce samples comparable to GANs on MNIST, CelebA and CIFAR-10 datasets, achieving a new state-of-the-art inception score of 8.87 on CIFAR-10. Additionally, we demonstrate that our models learn effective representations via image inpainting experiments.
Citations
Cited by
- Simultaneous Approximation of the Score Function and Its Derivatives by Deep Neural Networks
- Iterative Inference-time Scaling with Adaptive Frequency Steering for Image Super-Resolution
- Deterministic Image-to-Image Translation via Denoising Brownian Bridge Models with Dual Approximators
- SURE Guided Posterior Sampling: Trajectory Correction for Diffusion-Based Inverse Problems
- JADAI: Jointly Amortizing Adaptive Design and Bayesian Inference
- A Generative Reconstruction of Low-ℓ CMB B-Mode Signal using Reverse Diffusion in Deep Learning
- Energy-Guided Flow Matching Enables Few-Step Conformer Generation and Ground-State Identification
- Self-Evaluation Unlocks Any-Step Text-to-Image Generation
- Sampling with Shielded Langevin Monte Carlo Using Navigation Potentials
- Reverse-Time Diffusion Processes for Discrete Time Linear and Nonlinear Systems with non-Gaussian Noise
- Restoration Flow Matching-Based Channel Refinement and Equalization Correction for MIMO Semantic Communications
- Joint Activity Detection and Channel Estimation for Massive Connectivity: Where Message Passing Meets Score-Based Generative Priors
- Exploration of the generative capabilities of Boltzmann machines applied to social systems under the majority rule
- Diffusion-Guided Search via Exponential Tilting (DiffTilt): An Application to Falsification of Safety-Critical Systems
- From Score Learning to Discretized Sampling: An End-to-End Generalization Analysis of Diffusion Models
- Decompose and Reorganize: Planning with Primitives and Visuomotor Policies Learned from Demonstrations
- ScaleResfusion: Residual Rectified Flow based on Residual Vector Field
- Generative Distributionally Robust Optimization
- Accelerating Frequency Domain Diffusion Models with Error-Feedback Event-Driven Caching
- Ordering-based Causal Discovery via Generalized Score Matching
- DPAR: Dynamic Patchification for Efficient Autoregressive Visual Generation
- GriDiT: Factorized Grid-Based Diffusion for Efficient Long Image Sequence Generation
- Efficient and Robust Video Defense Framework against 3D-field Personalized Talking Face
- Generative Spectrum Cartography: Unified Reconstruction and Active Sensing via Diffusion Models
- LiDARDraft: Generating LiDAR Point Cloud from Versatile Inputs
- Spectral Diffusion for Sampling on \rm SU(N)
- Patlak Parametric Image Estimation from Dynamic PET Using Diffusion Model Prior
- The Ensemble Schrödinger Bridge filter for Nonlinear Data Assimilation
- MixFlow Training: Alleviating Exposure Bias with Slowed Interpolation Mixture
- In-Context Audio Control of Video Diffusion Transformers
- Is Your Conditional Diffusion Model Actually Denoising?
- Is There a Better Source Distribution than Gaussian? Exploring Source Distributions for Image Flow Matching
- PSI3D: Plug-and-Play 3D Stochastic Inference with Slice-wise Latent Diffusion Prior
- Efficient Zero-Shot Inpainting with Decoupled Diffusion Guidance
- Generative modeling of conditional probability distributions on the level-sets of collective variables
- Instant Expressive Gaussian Head Avatars at Over 100 FPS
- InfoDCL: Informative Noise Enhanced Diffusion Based Contrastive Learning
- Pixel Super-Resolved Fluorescence Lifetime Imaging Using Deep Learning
- SoFlow: Solution Flow Models for One-Step Generative Modeling
- Expand and Prune: Maximizing Trajectory Diversity for Effective GRPO in Generative Models
- EMFusion: Uncertainty-Aware Conditional Diffusion Model for Multivariate Narrow-band Exposure Forecasting
- Semantic Mismatch and Perceptual Degradation: A New Perspective on Image Editing Immunity
- Score-Based Turbo Message Passing for Plug-and-Play Compressive Imaging
- Structure From Tracking: Distilling Structure-Preserving Motion for Video Generation
- Coarse-to-Fine Hierarchical Alignment for UAV-based Human Detection using Diffusion Models
- Image Diffusion Preview with Consistency Solver
- Schrodinger Audio-Visual Editor: Object-Level Audiovisual Removal
- Multi-Robot Motion Planning from Vision and Language using Heat-Inspired Diffusion
- On Approaches to Building Surrogate ODE Models for Diffusion Bridges
- PIS: A Generalized Physical Inversion Solver for Arbitrary Sparse Observations via Set Conditioned Flow Matching
- Iterative Sampling Methods for Sinkhorn Distributionally Robust Optimization
- A probabilistic framework for crystal structure denoising, phase classification, and order parameters
- UNAAGI: Atom-Level Diffusion for Generating Non-Canonical Amino Acid Substitutions
- Robustness of Probabilistic Models to Low-Quality Data: A Multi-Perspective Analysis
- Diffusion Posterior Sampler for Hyperspectral Unmixing with Spectral Variability Modeling
- Gradient-Informed Monte Carlo Fine-Tuning of Diffusion Models for Low-Thrust Trajectory Design
- Physics-Guided Diffusion Priors for Multi-Slice Reconstruction in Scientific Imaging
- JoPano: Unified Panorama Generation via Joint Modeling
- PMA-Diffusion: A Physics-guided Mask-Aware Diffusion Framework for TSE from Sparse Observations
- Fast and Robust Diffusion Posterior Sampling for MR Image Reconstruction Using the Preconditioned Unadjusted Langevin Algorithm
- BulletTime: Decoupled Control of Time and Camera Pose for Video Generation
- Convergence Dynamics of Over-Parameterized Score Matching for a Single Gaussian
- Complementary Characterization of Agent-Based Models via Computational Mechanics and Diffusion Models
- Live Avatar: Streaming Real-time Audio-Driven Avatar Generation with Infinite Length
- ReflexFlow: Rethinking Learning Objective for Exposure Bias Alleviation in Flow Matching
- Data-regularized Reinforcement Learning for Diffusion Models at Scale
- On the Design of One-step Diffusion via Shortcutting Flow Paths
- Dimension-free error estimate for diffusion model and optimal scheduling
- SMP: Reusable Score-Matching Motion Priors for Physics-Based Character Control
- Glance: Accelerating Diffusion Models with 1 Sample
- From Navigation to Refinement: Revealing the Two-Stage Nature of Flow-based Diffusion Models through Oracle Velocity
- Diffusion-Prior Split Gibbs Sampling for Synthetic Aperture Radar Imaging under Incomplete Measurements
- Iterative Tilting for Diffusion Fine-Tuning
- Spatiotemporal Pyramid Flow Matching for Climate Emulation
- Improved Mean Flows: On the Challenges of Fastforward Generative Models
- Syndrome-Flow Consistency Model Achieves One-step Denoising Error Correction Codes
- Reversible Inversion for Training-Free Exemplar-guided Image Editing
- Adaptive-lambda Subtracted Importance Sampled Scores in Machine Unlearning for DDPMs and VAEs
- LAP: Fast LAtent Diffusion Planner for Autonomous Driving
- Overcoming the Curvature Bottleneck in MeanFlow
- Saddle-Free Guidance: Improved On-Manifold Sampling without Labels or Additional Training
- Generative Anchored Fields: Controlled Data Generation via Emergent Velocity Fields and Transport Algebra
- Generative models for crystalline materials
- Flow Density Control: Generative Optimization Beyond Entropy-Regularized Fine-Tuning
- MeanFlow Transformers with Representation Autoencoders
- TSGM: Regular and Irregular Time-series Generation using Score-based Generative Models
- LaGen: Towards Autoregressive LiDAR Scene Generation
- Deep Parameter Interpolation for Scalar Conditioning
- HMARK: Radioactive Multi-Bit Semantic-Latent Watermarking for Diffusion Models
- Flash-DMD: Towards High-Fidelity Few-Step Image Generation with Efficient Distillation and Joint Reinforcement Learning
- Learning to Generate Human-Human-Object Interactions from Textual Descriptions
- Back to the Feature: Explaining Video Classifiers with Video Counterfactual Explanations
- Advancing Image Classification with Discrete Diffusion Classification Modeling
- PRADA: Probability-Ratio-Based Attribution and Detection of Autoregressive-Generated Images
- Solving Diffusion Inverse Problems with Restart Posterior Sampling
- Flow Map Distillation Without Data
- Predicting partially observable dynamical systems via diffusion models with a multiscale inference scheme
- Understanding, Accelerating, and Improving MeanFlow Training
- ProxT2I: Efficient Reward-Guided Text-to-Image Generation via Proximal Diffusion
- Demystifying Diffusion Objectives: Reweighted Losses are Better Variational Bounds
- Breaking the Likelihood-Quality Trade-off in Diffusion Models by Merging Pretrained Experts
- CoD: A Diffusion Foundation Model for Image Compression
- Jacobian-Aware Posterior Sampling for Inverse Problems
- Sequence-Adaptive Video Prediction in Continuous Streams using Diffusion Noise Optimization
- WavefrontDiffusion: Dynamic Decoding Schedule for Improved Reasoning
- REXO: Indoor Multi-View Radar Object Detection via 3D Bounding Box Diffusion
- EvDiff: High Quality Video with an Event Camera
- DAPS++: Rethinking Diffusion Inverse Problems with Decoupled Posterior Annealing
- Diffusion-Inversion-Net (DIN): An End-to-End Direct Probabilistic Framework for Characterizing Hydraulic Conductivities and Quantifying Uncertainty
- Warm Diffusion: Recipe for Blur-Noise Mixture Diffusion Models
- SceneDesigner: Controllable Multi-Object Image Generation with 9-DoF Pose Manipulation
- MRI Super-Resolution with Deep Learning: A Comprehensive Survey
- TFCDiff: Robust ECG Denoising via Time-Frequency Complementary Diffusion
- AMS-KV: Adaptive KV Caching in Multi-Scale Visual Autoregressive Transformers
- Theoretical Closed-loop Stability Bounds for Dynamical System Coupled with Diffusion Policies
- Masked Auto-Regressive Variational Acceleration: Fast Inference Makes Practical Reinforcement Learning
- Efficient Score Pre-computation for Diffusion Models via Cross-Matrix Krylov Projection
- InstantViR: Real-Time Video Inverse Problem Solver with Distilled Diffusion Prior
- Energy-based Autoregressive Generation for Neural Population Dynamics
- CFG-EC: Error Correction Classifier-Free Guidance
- Scientific Data Compression and Super-Resolution Sampling
- Functional Mean Flow in Hilbert Space
- An unsupervised posterior sampling framework for multi-purpose seismic data recovery
- Scalable Simulation-Based Model Inference with Test-Time Complexity Control
- ActiveGrasp: Information-Guided Active Grasping with Calibrated Energy-based Model
- Stabilizing Self-Consuming Diffusion Models with Latent Space Filtering
- CaloClouds3: Ultra-Fast Geometry-Independent Highly-Granular Calorimeter Simulation
- PID-controlled Langevin Dynamics for Faster Sampling of Generative Models
- Diffusion Model Based Signal Recovery Under 1-Bit Quantization
- MFI-ResNet: Efficient ResNet Architecture Optimization via MeanFlow Compression and Selective Incubation
- PipeDiT: Accelerating Diffusion Transformers in Video Generation with Task Pipelining and Model Decoupling
- VPHO: Joint Visual-Physical Cue Learning and Aggregation for Hand-Object Pose Estimation
- SP-Guard: Selective Prompt-adaptive Guidance for Safe Text-to-Image Generation
- DenoGrad: A Gradient-Based Framework for Data Refinement in Tabular and Time-Series Learning
- Optimizing Input of Denoising Score Matching is Biased Towards Higher Score Norm
- Equivariant Sampling for Improving Diffusion Model-based Image Restoration
- Enhanced Privacy Leakage from Noise-Perturbed Gradients via Gradient-Guided Conditional Diffusion Models
- Diffusion Policies with Value-Conditional Optimization for Offline Reinforcement Learning
- Generative AI Meets 6G and Beyond: Diffusion Models for Semantic Communications
- Parallel Sampling via Autospeculation
- Rectified Noise: A Generative Model Using Positive-incentive Noise
- Forecasting implied volatility surface with generative diffusion models
- PADiff: Predictive and Adaptive Diffusion Policies for Ad Hoc Teamwork
- Laplacian Score Sharpening for Mitigating Hallucination in Diffusion Models
- Integrating Reweighted Least Squares with Plug-and-Play Diffusion Priors for Noisy Image Restoration
- GNN-Enabled Robust Hybrid Beamforming with Score-Based CSI Generation and Denoising
- Neodragon: Mobile Video Generation using Diffusion Transformer
- Joint Design of Protein Surface and Structure Using a Diffusion Bridge Model
- A PDE Perspective on Generative Diffusion Models
- DiScoFormer: Plug-In Density and Score Estimation with Transformers
- Enhancing Diffusion Model Guidance through Calibration and Regularization
- Integrating Score-Based Diffusion Models with Machine Learning-Enhanced Localization for Advanced Data Assimilation in Geological Carbon Storage
- Multi-agent Coordination via Flow Matching
- Sublinear iterations can suffice even for DDPMs
- Blind Strong Gravitational Lensing Inversion: Joint Inference of Source and Lens Mass with Score-Based Models
- Efficient probabilistic surrogate modeling techniques for partially-observed large-scale dynamical systems
- Score-Based Quickest Change Detection and Fault Identification for Multi-Stream Signals
- Conditional Score Learning for Quickest Change Detection in Markov Transition Kernels
- Robust inference using density-powered Stein operators
- Integrating Score-Based Generative Modeling and Neural ODEs for Accurate Representation of Multiscale Chaotic Dynamics
- FP-AbDiff: Improving Score-based Antibody Design by Capturing Nonequilibrium Dynamics through the Underlying Fokker-Planck Equation
- Scaling Multi-Agent Environment Co-Design with Diffusion Models
- A Conditional Diffusion Model for Building Energy Modeling Workflows
- Stability of mixed-state phases under weak decoherence
- Wasserstein Convergence of Critically Damped Langevin Diffusions
- Fractional Diffusion Bridge Models
- Diffusion Models at the Drug Discovery Frontier: A Review on Generating Small Molecules versus Therapeutic Peptides
- Optimal Convergence Analysis of DDPM for General Distributions
- Blind MIMO Semantic Communication via Parallel Variational Diffusion: A Completely Pilot-Free Approach
- Masked Diffusion Captioning for Visual Feature Learning
- Group-Equivariant Diffusion Models for Lattice Field Theory
- A Review of AI-Driven Approaches for Nanoscale Heat Conduction and Radiation
- On the Dataless Training of Neural Networks
- Physics-Guided Conditional Diffusion Networks for Microwave Image Reconstruction
- Off-policy Reinforcement Learning with Model-based Exploration Augmentation
- A Priori Sampling of Transition States with Guided Diffusion
- Denoising growth complexity: Data geometry and certified schedules for diffusion sampling
- Amortized Moment Matching for Visual Generation
- The Score Hamiltonian: Mapping Diffusion Models to Adiabatic Transport
- D2 Actor Critic: Diffusion Actor Meets Distributional Critic
- Vector-quantized Image Modeling with Improved VQGAN
- Beyond Prompts: Unconditional 3D Inversion for Out-of-Distribution Shapes
- DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation
- Diffusion Models in Vision: A Survey
- Anytime Sampling for Autoregressive Models via Ordered Autoencoding
- Sparse Transformer Architectures via Regularized Wasserstein Proximal Operator with L1 Prior
- Multi-Agent Scenario Generation in Roundabouts with a Transformer-enhanced Conditional Variational Autoencoder
- Causal Ordering for Structure Learning From Time Series
- Time-Embedded Algorithm Unrolling for Computational MRI
- ETC: training-free diffusion models acceleration with Error-aware Trend Consistency
- Information-Theoretic Discrete Diffusion
- Training-Free Safe Text Embedding Guidance for Text-to-Image Diffusion Models
- Score-based constrained generative modeling via Langevin diffusions with boundary conditions
- An efficient probabilistic hardware architecture for diffusion-like models
- LLM Meets Diffusion: A Hybrid Framework for Crystal Material Generation
- On the Anisotropy of Score-Based Generative Models
- Encoder-Decoder Diffusion Language Models for Efficient Training and Inference
- Distributionally Robust Optimization via Diffusion Ambiguity Modeling
- Optimize Any Topology: A Foundation Model for Shape- and Resolution-Free Structural Topology Optimization
- CANDI: Hybrid Discrete-Continuous Diffusion Models
- Learning density ratios in causal inference using Bregman-Riesz regression
- DiffusionLane: Diffusion Model for Lane Detection
- Diffusion-Driven Two-Stage Active Learning for Low-Budget Semantic Segmentation
- Scaling Non-Parametric Sampling with Representation
- Learning 3D Anisotropic Noise Distributions Improves Molecular Force Field Modeling
- Generalised Flow Maps for Few-Step Generative Modelling on Riemannian Manifolds
- SCORENF: Score-based Normalizing Flows for Sampling Unnormalized distributions
- Fisher meets Feynman: score-based variational inference with a product of experts
- Generative AI in Depth: A Survey of Recent Advances, Model Variants, and Real-World Applications
- Generative Point Tracking with Flow Matching
- BadGraph: A Backdoor Attack Against Latent Diffusion Model for Text-Guided Graph Generation
- Exponential Convergence Guarantees for Iterative Markovian Fitting
- Information Gradient for Nonlinear Gaussian Channel with Applications to Task-Oriented Communication
- AlphaFlow: Understanding and Improving MeanFlow Models
- UniHPR: Unified Human Pose Representation via Singular Value Contrastive Learning
- AI-Aided Annealed Langevin Dynamics for Rapid Optimization of Programmable Channels
- Quantum Reversibility Meets Classical Reverse Diffusion
- DiffGRM: Diffusion-based Generative Recommendation Model
- GOOD: Training-Free Guided Diffusion Sampling for Out-of-Distribution Detection
- Matricial Free Energy as a Gaussianizing Regularizer: Enhancing Autoencoders for Gaussian Code Generation
- KineDiff3D: Kinematic-Aware Diffusion for Category-Level Articulated Object Shape Reconstruction and Generation
- Disentanglement of Sources in a Multi-Stream Variational Autoencoder
- Particle Dynamics for Latent-Variable Energy-Based Models
- Operator Flow Matching for Timeseries Forecasting
- Fast and Scalable Score-Based Kernel Calibration Tests
- Adapting Self-Supervised Representations as a Latent Space for Efficient Generation
- Exploring Cross-Modal Flows for Few-Shot Learning
- Inference-Time Search using Side Information for Diffusion-based Image Reconstruction
- pi-Flow: Policy-Based Few-Step Generation via Imitation Distillation
- Briding Diffusion Posterior Sampling and Monte Carlo methods: a survey
- PriorGuide: Test-Time Prior Adaptation for Simulation-Based Inference
- GO-Diff: Data-free and amortized global structure optimization
- Adaptive Visual Conditioning for Semantic Consistency in Diffusion-Based Story Continuation
- Kinematic Kitbashing for Modeling Functional Articulated Objects
- A Connection Between Score Matching and Local Intrinsic Dimension
- RNAGenScape: Property-guided Optimization and Interpolation of mRNA Sequences with Manifold Langevin Dynamics
- DiffEM: Learning from Corrupted Data with Diffusion Models via Expectation Maximization
- Adapting Noise to Data: Generative Flows from 1D Processes
- LayerSync: Self-aligning Intermediate Layers
- Mitigating the Noise Shift for Denoising Generative Models via Noise Awareness Guidance
- Improving Generative Behavior Cloning via Self-Guidance and Adaptive Chunking
- Scene Coordinate Reconstruction Priors
- Learning Latent Energy-Based Models via Interacting Particle Langevin Dynamics
- MosaicDiff: Training-free Structural Pruning for Diffusion Model Acceleration Reflecting Pretraining Dynamics
- Diffusion-DFL: Decision-focused Diffusion Models for Stochastic Optimization
- Offline Reinforcement Learning with Generative Trajectory Policies
- Schrödinger bridge for generative AI: Soft-constrained formulation and convergence analysis
- Discrete State Diffusion Models: A Sample Complexity Perspective
- Fine-Tuning Flow Matching via Maximum Likelihood Estimation of Reconstructions
- Using neural style transfer to study the evolution of animal signal design: A case study in an ornamented fish
- Asymmetric Flow Models
- Learning Regularization Functionals for Inverse Problems: A Comparative Study
- The turn less taken: Investigating patterns in <i>β</i> -turn dynamics using large-scale molecular dynamics data
- MCMC: Bridging Rendering, Optimization and Generative AI
- Score-Based Density Estimation from Pairwise Comparisons
- SummDiff: Generative Modeling of Video Summarization with Diffusion
- Guided Star-Shaped Masked Diffusion
- Ctrl-VI: Controllable Video Synthesis via Variational Inference
- Deep Neural Networks Inspired by Differential Equations
- Wavefunction Flows: Efficient Quantum Simulation of Continuous Flow Models
- RePainter: Empowering E-commerce Object Removal via Spatial-matting Reinforcement Learning
- Geometry-aware Policy Imitation
- Large Scale Diffusion Distillation via Score-Regularized Continuous-Time Consistency
- SkipSR: Faster Super Resolution with Token Skipping
- Next Semantic Scale Prediction via Hierarchical Diffusion Language Models
- EigenScore: OOD Detection using Covariance in Diffusion Models
- MV-Performer: Taming Video Diffusion Model for Faithful and Synchronized Multi-view Performer Synthesis
- An Inertial Langevin Algorithm
- Rethinking Nonlinearity: Trainable Gaussian Mixture Modules for Modern Neural Architectures
- Diffusion Models and the Manifold Hypothesis: Log-Domain Smoothing is Geometry Adaptive
- Control-Augmented Autoregressive Diffusion for Data Assimilation
- Carré du champ flow matching: better quality-generalisation tradeoff in generative models
- Efficient Conditional Generation on Scale-based Visual Autoregressive Models
- Mutual Information Estimation via Score-to-Fisher Bridge for Nonlinear Gaussian Noise Channels
- A Data-Driven Prism: Multi-View Source Separation with Diffusion Model Priors
- Policy Gradient Guidance Enables Test Time Control
- SSDD: Single-Step Diffusion Decoder for Efficient Image Tokenization
- Improved probabilistic regression using diffusion models
- Language Model Based Text-to-Audio Generation: Anti-Causally Aligned Collaborative Residual Transformers
- TAG:Tangential Amplifying Guidance for Hallucination-Resistant Diffusion Sampling
- MACS: Measurement-Aware Consistency Sampling for Inverse Problems
- Mitigating Diffusion Model Hallucinations with Dynamic Guidance
- Joint Learning of Pose Regression and Denoising Diffusion with Score Scaling Sampling for Category-level 6D Pose Estimation
- Exploring the Power of Diffusion Large Language Models for Software Engineering: An Empirical Investigation
- A Diffusion-based Generative Machine Learning Paradigm for Dynamic Contingency Screening
- Self-supervised diffusion model fine-tuning for costate initialization using Markov chain Monte Carlo
- Unlearning in Diffusion models under Data Constraints: A Variational Inference Approach
- Optimal estimation of a factorizable density using diffusion models with ReLU neural networks
- Flexible Locomotion Learning with Diffusion Model Predictive Control
- Diffusion2: Dual Diffusion Model with Uncertainty-Aware Adaptive Noise for Momentary Trajectory Prediction
- A Novel Cloud-Based Diffusion-Guided Hybrid Model for High-Accuracy Accident Detection in Intelligent Transportation Systems
- Learning Robust Diffusion Models from Imprecise Supervision
- Dale meets Langevin: A Multiplicative Denoising Diffusion Model
- LVTINO: LAtent Video consisTency INverse sOlver for High Definition Video Restoration
- Combining complex Langevin dynamics with score-based and energy-based diffusion models
- Syntax-Guided Diffusion Language Models with User-Integrated Personalization
- Riemannian Consistency Model
- Plug-and-Play Prompt Refinement via Latent Feedback for Diffusion Model Alignment
- CINDES: Classification induced neural density estimator and simulator
- A Geometric Unification of Generative AI with Manifold-Probabilistic Projection Models
- Learn to Guide Your Diffusion Model
- DiffAU: Diffusion-Based Ambisonics Upscaling
- EnScale: Temporally-consistent multivariate generative downscaling via proper scoring rules
- EVODiff: Entropy-aware Variance Optimized Diffusion Inference
- Fading to Grow: Growing Preference Ratios via Preference Fading Discrete Diffusion for Recommendation
- Score Distillation of Flow Matching Models
- Score-based Membership Inference on Diffusion Models
- When Scores Learn Geometry: Rate Separations under the Manifold Hypothesis
- Assessing the risk of future Dunkelflaute events for Germany using generative deep learning
- CMT: Mid-Training for Efficient Learning of Consistency, Mean Flow, and Flow Map Models
- ASIA: Adaptive 3D Segmentation using Few Image Annotations
- A phase transition in diffusion models reveals the hierarchical nature of data
- Diffusion Schr "odinger Bridge with Applications to Score-Based\n Generative Modeling
- NeoWorld: Neural Simulation of Explorable Virtual Worlds via Progressive 3D Unfolding
- StrucADT: Generating Structure-controlled 3D Point Clouds with Adjacency Diffusion Transformer
- Avoid Catastrophic Forgetting with Rank-1 Fisher from Diffusion Models
- Transport Based Mean Flows for Generative Modeling
- HiGS: History-Guided Sampling for Plug-and-Play Enhancement of Diffusion Models
- Factor-Based Conditional Diffusion Model for Portfolio Optimization
- High-Quality Sound Separation Across Diverse Categories via Visually-Guided Generative Modeling
- Universal Multi-Domain Translation via Diffusion Routers
- FlowDrive: moderated flow matching with data balancing for trajectory planning
- Noise-to-Notes: Diffusion-based Generation and Refinement for Automatic Drum Transcription
- A Unifying Framework for Parallelizing Sequential Models with Linear Dynamical Systems
- Generating Stable Placements via Physics-guided Diffusion Models
- DriftLite: Lightweight Drift Control for Inference-Time Scaling of Diffusion Models
- Responsible Diffusion: A Comprehensive Survey on Safety, Ethics, and Trust in Diffusion Models
- Causal Time Series Generation via Diffusion Models
- PhysCtrl: Generative Physics for Controllable and Physics-Grounded Video Generation
- Incomplete Data, Complete Dynamics: A Diffusion Approach
- Diffusion and Flow-based Copulas: Forgetting and Remembering Dependencies
- A Recovery Theory for Diffusion Priors: Deterministic Analysis of the Implicit Prior Algorithm
- Frame-based Equivariant Diffusion Models for 3D Molecular Generation
- From Synthetic to Real: Toward Identity-Consistent Makeup Transfer with Synthetic and Real Data
- Rao-Blackwellized Score Matching on Manifolds
- Trust Region Inverse Reinforcement Learning: Explicit Dual Ascent using Local Policy Updates
- Harnessing Synthetic Data from Generative AI for Statistical Inference
- VARS-fUSI: Variable Sampling for Fast and Efficient Functional Ultrasound Imaging using Neural Operators
- Generalizability of Predictive and Generative Speech Enhancement Models to Pathological Speakers
- Influence of Clean Speech Characteristics on Speech Enhancement Performance
- On the Edge of Memorization in Diffusion Models
- CAR-Flow: Condition-Aware Reparameterization Aligns Source and Target for Better Flow Matching
- HOGraspFlow: Exploring Vision-based Generative Grasp Synthesis with Hand-Object Priors and Taxonomy Awareness
- Density Ratio Estimation via Infinitesimal Classification
- A Closer Look at Model Collapse: From a Generalization-to-Memorization Perspective
- Factorizing Diffusion Policies for Observation Modality Prioritization
- UniTac2Pose: A Unified Approach Learned in Simulation for Category-level Visuotactile In-hand Pose Estimation
- Latent Zoning Network: A Unified Principle for Generative Modeling, Representation Learning, and Classification
- Local Mechanisms of Compositional Generalization in Conditional Diffusion
- Kuramoto Orientation Diffusion Models
- Lightweight and Accurate Multi-View Stereo with Confidence-Aware Diffusion Model
- Quantum neural ordinary and partial differential equations
- Optimizing Multi-Modal Trackers via Sensitivity-aware Regularized Tuning
- Variational Shape Inference for Grasp Diffusion on SE(3)
- RespoDiff: Dual-Module Bottleneck Transformation for Responsible & Faithful T2I Generation
- Towards Robust Defense against Customization via Protective Perturbation Resistant to Diffusion-based Purification
- Inverse Design of Amorphous Materials with Targeted Properties
- DiCache: Let Diffusion Model Determine Its Own Cache
- BWCache: Accelerating Video Diffusion Transformers through Block-Wise Caching
- SpeechOp: Inference-Time Task Composition for Generative Speech Processing
- Cross-Distribution Diffusion Priors-Driven Iterative Reconstruction for Sparse-View CT
- Generalizable Holographic Reconstruction via Amplitude-Only Diffusion Priors
- Diffusion Models Beat GANs on Image Synthesis
- Adaptive Sampling Scheduler
- A Statistical Benchmark for Diffusion Posterior Sampling Algorithms
- HoloGarment: 360° Novel View Synthesis of In-the-Wild Garments
- Solving ill-conditioned polynomial equations using score-based priors with application to multi-target detection
- Beyond Sliders: Mastering the Art of Diffusion-based Image Manipulation
- AI-Generated Content in Cross-Domain Applications: Research Trends, Challenges and Propositions
- A Survey of Deep Learning-based Point Cloud Denoising
- Score Matching on Large Geometric Graphs for Cosmology Generation
- Automated Tuning for Diffusion Inverse Problem Solvers without Generative Prior Retraining
- A novel method and dataset for depth-guided image deblurring from smartphone Lidar
- LatentVoiceGrad: Nonparallel Voice Conversion with Latent Diffusion/Flow-Matching Models
- MV-RAG: Retrieval Augmented Multiview Diffusion
- Physics-Guided Null-Space Diffusion with Sparse Masking for Corrective Sparse-View CT Reconstruction
- Transition Models: Rethinking the Generative Learning Objective
- Hyper Diffusion Avatars: Dynamic Human Avatar Generation using Network Weight Space Diffusion
- Wavelet Fourier Diffuser: Frequency-Aware Diffusion Model for Reinforcement Learning
- Divergence-kernel method for linear responses of densities and generative models
- Diffusion Generative Models Meet Compressed Sensing, with Applications to Imaging and Finance
- On the MIA Vulnerability Gap Between Private GANs and Diffusion Models
- Scale-Adaptive Generative Flows for Multiscale Scientific Data
- Generative AI for Crystal Structures: A Review
- A Sharp KL-Convergence Analysis for Diffusion Models under Minimal Assumptions
- OmniCache: A Trajectory-Oriented Global Perspective on Training-Free Cache Reuse for Diffusion Transformer Models
- On the Collapse Errors Induced by the Deterministic Sampler for Diffusion Models
- Lipschitz-Guided Design of Interpolation Schedules in Generative Models
- A Continuous-Time Consistency Model for 3D Point Cloud Generation
- C-DiffDet+: Fusing Global Scene Context with Generative Denoising for High-Fidelity Car Damage Detection
- Generative Latent Space Dynamics of Electron Density
- Weighted Support Points from Random Measures: An Interpretable Alternative for Generative Modeling
- VarDiU: A Variational Diffusive Upper Bound for One-Step Diffusion Distillation
- Train-Once Plan-Anywhere Kinodynamic Motion Planning via Diffusion Trees
- Guiding Noisy Label Conditional Diffusion Models with Score-based Discriminator Correction
- Towards 6G Intelligence: The Role of Generative AI in Future Wireless Networks
- Reduced-Order Modeling of Cyclo-Stationary Time Series Using Score-Based Generative Methods
- STDiff: A State Transition Diffusion Framework for Time Series Imputation in Industrial Systems
- Energy-Based Flow Matching for Generating 3D Molecular Structure
- Hybrid Perception and Equivariant Diffusion for Robust Multi-Node Rebar Tying
- Parallelizing MCMC Across the Sequence Length
- Incorporating Pre-trained Diffusion Models in Solving the Schrödinger Bridge Problem
- GSFix3D: Diffusion-Guided Repair of Novel Views in Gaussian Splatting
- CTA-Flux: Integrating Chinese Cultural Semantics into High-Quality English Text-to-Image Communities
- A Non-Asymptotic Convergent Analysis for Scored-Based Graph Generative Model via a System of Stochastic Differential Equations
- FOCUS: Frequency-Optimized Conditioning of DiffUSion Models for mitigating catastrophic forgetting during Test-Time Adaptation
- SBGD: Improving Graph Diffusion Generative Model via Stochastic Block Diffusion
- Sampling by averaging: A multiscale approach to score estimation
- A Convergent Primal-Dual Algorithm for Computing Rate-Distortion-Perception Functions
- One-Step Flow Q-Learning: Addressing the Diffusion Policy Bottleneck in Offline Reinforcement Learning
- Counterfactual Probabilistic Diffusion with Expert Models
- Reinforced Context Order Recovery for Adaptive Reasoning and Planning
- Fast Convergence for Langevin Diffusion with Manifold Structure
- Residual-based Efficient Bidirectional Diffusion Model for Image Dehazing and Haze Generation
- Probabilistic Forecasting Method for Offshore Wind Farm Cluster under Typhoon Conditions: a Score-Based Conditional Diffusion Model
- Energy-Based Models for Predicting Mutational Effects on Proteins
- Projected Coupled Diffusion for Test-Time Constrained Joint Generation
- Alternating Approach-Putt Models for Multi-Stage Speech Enhancement
- Translation of Text Embedding via Delta Vector to Suppress Strongly Entangled Content in Text-to-Image Diffusion Models
- Towards Spatially Consistent Image Generation: On Incorporating Intrinsic Scene Properties into Diffusion Models
- ReconVLA: Reconstructive Vision-Language-Action Model as Effective Robot Perceiver
- Non-asymptotic convergence bound of conditional diffusion models
- MUJICA: Reforming SISR Models for PBR Material Super-Resolution via Cross-Map Attention
- CrystalDiT: A Diffusion Transformer for Crystal Generation
- Images Speak Louder Than Scores: Failure Mode Escape for Enhancing Generative Quality
- Distilling LLM Prior to Flow Model for Generalizable Agent's Imagination in Object Goal Navigation
- Flow-SLM: Joint Learning of Linguistic and Acoustic Information for Spoken Language Modeling
- Hybrid Long and Short Range Flows for Point Cloud Filtering
- DiffVolume: Diffusion Models for Volume Generation in Limit Order Books
- Towards Robust Active Feature Acquisition
- Designing Object Detection Models for TinyML: Foundations, Comparative Analysis, Challenges, and Emerging Solutions
- An effective potential for generative modelling with active matter
- Score Augmentation for Diffusion Models
- Efficient Approximate Posterior Sampling with Annealed Langevin Monte Carlo
- AR-GRPO: Training Autoregressive Image Generation Models via Reinforcement Learning
- A Score-based Diffusion Model Approach for Adaptive Learning of Stochastic Partial Differential Equation Solutions
- Local Diffusion Models and Phases of Data Distributions
- Improved Sub-Visible Particle Classification in Flow Imaging Microscopy via Generative AI-Based Image Synthesis
- A Rose by Any Other Name Would Smell as Sweet: Categorical Homotopy Theory for Large Language Models
- Multitask Learning with Stochastic Interpolants
- Multi-Marginal Stochastic Flow Matching for High-Dimensional Snapshot Data at Irregular Time Points
- Turbulent Injection assisted by Diffusion Models for Scale Resolving Simulations
- DocVCE: Diffusion-based Visual Counterfactual Explanations for Document Image Classification
- Generating Feasible and Diverse Synthetic Populations Using Diffusion Models
- Diffusion Models with Adaptive Negative Sampling Without External Resources
- Convergence of Deterministic and Stochastic Diffusion-Model Samplers: A Simple Analysis in Wasserstein Distance
- Forecasting When to Forecast: Accelerating Diffusion Models with Confidence-Gated Taylor
- Diffusion models for inverse problems
- Scoring ISAC: Benchmarking Integrated Sensing and Communications via Score-Based Generative Modeling
- Seed Diffusion: A Large-Scale Diffusion Language Model with High-Speed Inference
- Why Heuristic Weighting Works: A Theoretical Analysis of Denoising Score Matching
- DisCo3D: Distilling Multi-View Consistency for 3D Scene Editing
- Diffusion Models for Future Networks and Communications: A Comprehensive Survey
- DPoser-X: Diffusion Model as Robust 3D Whole-body Human Pose Prior
- Weakly Supervised Virus Capsid Detection with Image-Level Annotations in Electron Microscopy Images
- Jet Image Generation in High Energy Physics Using Diffusion Models
- DiffuMatch: Category-Agnostic Spectral Diffusion Priors for Robust Non-rigid Shape Matching
- Stable-Sim2Real: Exploring Simulation of Real-Captured 3D Data with Two-Stage Depth Diffusion
- Set Invariance with Probability One for Controlled Diffusion: Score-based Approach
- MapDiffusion: Generative Diffusion for Vectorized Online HD Map Construction and Uncertainty Estimation in Autonomous Driving
- Flow Matching Policy Gradients
Discussions
Related