Research map / Probabilistic and scientific modeling
Variational and generative sampling: research map
2,060 accepted papers on Variational and generative sampling in Probabilistic and scientific modeling, from ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026), grouped into 6 clusters and 24 approaches. The busiest year so far is 2026.
Within Probabilistic and scientific modeling, its share shrank from 20.4% in 2023–24 to 16.4% in 2025–26 (414 → 499 papers at ICML, NeurIPS, CVPR and AAAI, the venues with data for all four years).
Explore Variational and generative sampling in the interactive map
Working on something in this topic? Describe your idea in scime atlas to see which approach it falls under, the closest papers by meaning and how crowded the spot has become.
Approaches and key papers
estimation · probabilistic · variables · 546 papers
Approaches in this cluster:
- Likelihood-free and graphical model estimation (161 papers)
Estimates intractable models using kernel ABC, Stein tests and sparse graphical models. - Tractable probabilistic models (90 papers)
Builds tractable density models and circuits that permit exact inference. - Exact inference for probabilistic programs (103 papers)
Performs exact or scalable inference in discrete and mixed probabilistic programs. - Latent variable marginalization and estimators (91 papers)
Estimates marginal likelihood and marginalizes discrete latent variables with unbiased estimators. - Gradient estimation for discrete variables (74 papers)
Backpropagates through discrete latents with straight-through, control-variate and perturbation estimators. - Mutual information estimation (27 papers)
Benchmarks and designs neural estimators of mutual information.
Most cited and most cited since 2024:
- Categorical Reparameterization with Gumbel-Softmax (ICLR 2017 · 3,310 citations)
- The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables (ICLR 2017 · 1,471 citations)
- Differentiable Information Bottleneck for Deterministic Multi-view Clustering (CVPR 2024 · 27 citations)
- Latent Modulated Function for Computational Optimal Continuous Image Representation (CVPR 2024 · 16 citations)
diffusion · score · matching · 409 papers
Approaches in this cluster:
- Posterior sampling with diffusion guidance (137 papers)
Samples posteriors or unnormalized targets using gradient flows, Stein guidance and amortized samplers. - Score-based generative modeling analysis (128 papers)
Estimates data gradients and analyzes bias, generalization and evaluation of generative models. - Flow matching and flow sampling (144 papers)
Learns straighter flows and samples from unnormalized densities via flow matching.
Most cited and most cited since 2024:
- Wasserstein Generative Adversarial Networks (ICML 2017 · 4,925 citations)
- Generative Modeling by Estimating Gradients of the Data Distribution (NeurIPS 2019 · 1,076 citations)
- Discrete Flow Matching (NeurIPS 2024 · 16 citations)
- Optimal Flow Matching: Learning Straight Trajectories in Just One Step (NeurIPS 2024 · 11 citations)
vae · autoencoders · latent · 365 papers
Approaches in this cluster:
- Structured latent VAEs (125 papers)
Improves VAEs with dependent, discrete and Laplace-based latent structure and generalization bounds. - Regularizing VAE latent information (89 papers)
Controls information flow in latents via deterministic autoencoders, hierarchy and divergence regularizers. - Disentangled and discrete latent models (92 papers)
Learns discrete or disentangled latent concepts with identifiability analysis. - Hierarchical VAEs and posterior collapse (59 papers)
Deepens latent hierarchies and prevents posterior collapse in autoencoder models.
Most cited and most cited since 2024:
- beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework (ICLR 2017 · 2,979 citations)
- Neural Discrete Representation Learning (NeurIPS 2017 · 2,816 citations)
- Robust Multimodal Survival Prediction with Conditional Latent Differentiation Variational AutoEncoder (CVPR 2025 · 13 citations)
- Learned Lossless Image Compression based on Bit Plane Slicing (CVPR 2024 · 11 citations)
monte · carlo · mcmc · 280 papers
Approaches in this cluster:
- Hamiltonian Monte Carlo (89 papers)
Improves Hamiltonian Monte Carlo through reflection, Riemannian geometry and neural parameterization. - Sequential Monte Carlo for inference (78 papers)
Uses twisted and learned sequential Monte Carlo for language-model and latent-variable inference. - Stochastic gradient MCMC (68 papers)
Samples large-scale distributions with stochastic gradient and pseudo-extended samplers. - Importance sampling and quadrature (45 papers)
Reduces estimator variance using orthogonal Monte Carlo, annealing and control variates.
Most cited and most cited since 2024:
- A Kernel Test of Goodness of Fit (ICML 2016 · 92 citations)
- Fast and Provably Good Seedings for k-Means (NeurIPS 2016 · 84 citations)
- Principled Probabilistic Imaging using Diffusion Models as Plug-and-Play Priors (NeurIPS 2024 · 11 citations)
- Using Stratified Sampling to Improve LIME Image Explanations (AAAI 2024 · 7 citations)
variational inference · vi · lower · 267 papers
Approaches in this cluster:
- Variational inference theory and gradients (89 papers)
Gives guarantees and lower-variance gradient estimators for variational inference. - Annealing and divergence-based variational objectives (79 papers)
Uses annealed importance sampling, alpha-divergences and thermodynamic objectives for variational bounds. - Amortized and online variational inference (59 papers)
Amortizes variational inference across tasks, filtering and undirected models. - Variational methods for sequence models (40 papers)
Applies variational inference to recurrent models, deep networks and GFlowNets.
Most cited and most cited since 2024:
- A Weighted Variational Model for Simultaneous Reflectance and Illumination Estimation (CVPR 2016 · 1,104 citations)
- Composing graphical models with neural networks for structured representations and fast inference (NeurIPS 2016 · 216 citations)
- Q-MAML: Quantum Model-Agnostic Meta-Learning for Variational Quantum Algorithms (AAAI 2025 · 7 citations)
- GAD-PVI: A General Accelerated Dynamic-Weight Particle-Based Variational Inference Framework (AAAI 2024 · 5 citations)
flows · normalizing · invertible · 193 papers
Approaches in this cluster:
- Normalizing flow training and design (88 papers)
Builds normalizing flows with stochastic interpolants, dense connections and symmetry equivariance. - Expressive invertible architectures (76 papers)
Improves flow expressiveness via variational dequantization and residual or Koopman-based designs. - Flow-based density estimation (29 papers)
Applies flows to density-ratio estimation and likelihood analysis with autoregressive structure.
Most cited and most cited since 2024:
- Improved Variational Inference with Inverse Autoregressive Flow (NeurIPS 2016 · 762 citations)
- Flow++: Improving Flow-Based Generative Models with Variational Dequantization and Architecture Design (ICML 2019 · 206 citations)
- Flow Matching on General Geometries (ICLR 2024 · 5 citations)
- An amortized approach to non-linear mixed-effects modeling based on neural posterior estimation (ICML 2024 · 4 citations)
Related topics in Probabilistic and scientific modeling
- Dynamical systems (969)
- Uncertainty and Gaussian processes (1,162)
- Molecules and proteins (1,990)
- Treatment effects (421)
- Differential equations and PDEs (614)
- Causal discovery (830)
- Bayesian optimization (397)