NeurIPS 2021 accepted papers on Probabilistic and scientific modeling
304 of the 2,508 papers accepted at NeurIPS 2021 (12.1%) are on Probabilistic and scientific modeling, compared with 14.2% at NeurIPS 2020. They fall into 8 topics.
See NeurIPS 2021 in the interactive map
Submitting to NeurIPS? Checklist →
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Variational and generative sampling · 82 papers
- Learning Generative Vision Transformer with Energy-Based Latent Space for Saliency Prediction (45 citations)
- Robust Compressed Sensing MRI with Deep Generative Priors (45 citations)
- Equilibrium and non-Equilibrium regimes in the learning of Restricted Boltzmann Machines (36 citations)
- Why Do Pretrained Language Models Help in Downstream Tasks? An Analysis of Head and Prompt Tuning (33 citations)
- Smooth Normalizing Flows (17 citations)
- Contrastively Disentangled Sequential Variational Autoencoder (14 citations)
- Densely connected normalizing flows (13 citations)
- End-to-end reconstruction meets data-driven regularization for inverse problems (12 citations)
- Identifiable Generative models for Missing Not at Random Data Imputation (12 citations)
- On Memorization in Probabilistic Deep Generative Models (12 citations)
- Score-based Generative Modeling in Latent Space (12 citations)
- Consistency Regularization for Variational Auto-Encoders (11 citations)
and 70 more in the interactive map.
Uncertainty and Gaussian processes · 57 papers
- Uncertainty Quantification and Deep Ensembles (31 citations)
- Laplace Redux - Effortless Bayesian Deep Learning (27 citations)
- Asymptotics of representation learning in finite Bayesian neural networks (24 citations)
- Reliable Post hoc Explanations: Modeling Uncertainty in Explainability (23 citations)
- Repulsive Deep Ensembles are Bayesian (16 citations)
- Beyond Pinball Loss: Quantile Methods for Calibrated Uncertainty Quantification (15 citations)
- Benchmarking Bayesian Deep Learning on Diabetic Retinopathy Detection Tasks (14 citations)
- Learning to Learn Dense Gaussian Processes for Few-Shot Learning (11 citations)
- Exact marginal prior distributions of finite Bayesian neural networks (10 citations)
- Variational Multi-Task Learning with Gumbel-Softmax Priors (9 citations)
- Scaling Ensemble Distribution Distillation to Many Classes with Proxy Targets (8 citations)
- Scaling Gaussian Processes with Derivative Information Using Variational Inference (8 citations)
and 45 more in the interactive map.
Molecules and proteins · 42 papers
- Language models enable zero-shot prediction of the effects of mutations on protein function (699 citations)
- FLIP: Benchmark tasks in fitness landscape inference for proteins (146 citations)
- Learning Graph Models for Retrosynthesis Prediction (71 citations)
- GeoMol: Torsional Geometric Generation of Molecular 3D Conformer Ensembles (52 citations)
- Reinforcement learning for optimization of variational quantum circuit architectures (51 citations)
- Explaining heterogeneity in medial entorhinal cortex with task-driven neural networks (38 citations)
- Flow Network based Generative Models for Non-Iterative Diverse Candidate Generation (28 citations)
- Adaptive Conformal Inference Under Distribution Shift (27 citations)
- Multi-Scale Representation Learning on Proteins (22 citations)
- Towards understanding retrosynthesis by energy-based models (17 citations)
- ATOM3D: Tasks on Molecules in Three Dimensions (16 citations)
- Deep Extended Hazard Models for Survival Analysis (16 citations)
and 30 more in the interactive map.
Dynamical systems · 40 papers
- Neural Latents Benchmark ‘21: Evaluating latent variable models of neural population activity (44 citations)
- Causal Navigation by Continuous-time Neural Networks (25 citations)
- Targeted Neural Dynamical Modeling (22 citations)
- Charting and Navigating the Space of Solutions for Recurrent Neural Networks (10 citations)
- Differentiable Simulation of Soft Multi-body Systems (10 citations)
- Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons (10 citations)
- A Bayesian-Symbolic Approach to Reasoning and Learning in Intuitive Physics (9 citations)
- A universal probabilistic spike count model reveals ongoing modulation of neural variability (9 citations)
- Impression learning: Online representation learning with synaptic plasticity (7 citations)
- Neural Hybrid Automata: Learning Dynamics With Multiple Modes and Stochastic Transitions (7 citations)
- Reverse engineering recurrent neural networks with Jacobian switching linear dynamical systems (7 citations)
- Accurately Solving Rod Dynamics with Graph Learning (6 citations)
and 28 more in the interactive map.
Causal discovery · 33 papers
- Self-Supervised Learning with Data Augmentations Provably Isolates Content from Style (69 citations)
- DECAF: Generating Fair Synthetic Data Using Causally-Aware Generative Networks (26 citations)
- A Causal Lens for Controllable Text Generation (19 citations)
- Recovering Latent Causal Factor for Generalization to Distributional Shifts (19 citations)
- The Causal-Neural Connection: Expressiveness, Learnability, and Inference (12 citations)
- Interventional Sum-Product Networks: Causal Inference with Tractable Probabilistic Models (11 citations)
- MIRACLE: Causally-Aware Imputation via Learning Missing Data Mechanisms (11 citations)
- Beware of the Simulated DAG! Causal Discovery Benchmarks May Be Easy to Game (10 citations)
- Nested Counterfactual Identification from Arbitrary Surrogate Experiments (8 citations)
- Higher Order Kernel Mean Embeddings to Capture Filtrations of Stochastic Processes (7 citations)
- Dynamic Causal Bayesian Optimization (6 citations)
- A Critical Look at the Consistency of Causal Estimation with Deep Latent Variable Models (5 citations)
and 21 more in the interactive map.
Treatment effects · 19 papers
- On Inductive Biases for Heterogeneous Treatment Effect Estimation (11 citations)
- SyncTwin: Treatment Effect Estimation with Longitudinal Outcomes (10 citations)
- A/B Testing for Recommender Systems in a Two-sided Marketplace (9 citations)
- Causal-BALD: Deep Bayesian Active Learning of Outcomes to Infer Treatment-Effects from Observational Data (9 citations)
- Causal Effect Inference for Structured Treatments (8 citations)
- Double/Debiased Machine Learning for Dynamic Treatment Effects (8 citations)
- Really Doing Great at Estimating CATE? A Critical Look at ML Benchmarking Practices in Treatment Effect Estimation (8 citations)
- SurvITE: Learning Heterogeneous Treatment Effects from Time-to-Event Data (7 citations)
- Identification and Estimation of Joint Probabilities of Potential Outcomes in Observational Studies with Covariate Information (6 citations)
- Estimating Multi-cause Treatment Effects via Single-cause Perturbation (5 citations)
- Learning Treatment Effects in Panels with General Intervention Patterns (5 citations)
- Double Machine Learning Density Estimation for Local Treatment Effects with Instruments (3 citations)
and 7 more in the interactive map.
Differential equations and PDEs · 19 papers
- Characterizing possible failure modes in physics-informed neural networks (118 citations)
- Multiwavelet-based Operator Learning for Differential Equations (64 citations)
- Integrating Expert ODEs into Neural ODEs: Pharmacology and Disease Progression (20 citations)
- Heavy Ball Neural Ordinary Differential Equations (15 citations)
- On the Validity of Modeling SGD with Stochastic Differential Equations (SDEs) (15 citations)
- NTopo: Mesh-free Topology Optimization using Implicit Neural Representations (12 citations)
- Neural Symplectic Form: Learning Hamiltonian Equations on General Coordinate Systems (7 citations)
- Conditionally Parameterized, Discretization-Aware Neural Networks for Mesh-Based Modeling of Physical Systems (5 citations)
- Sparse Flows: Pruning Continuous-depth Models (4 citations)
- A Probabilistic State Space Model for Joint Inference from Differential Equations and Data (3 citations)
- Second-Order Neural ODE Optimizer (3 citations)
- Scalable Inference in SDEs by Direct Matching of the Fokker–Planck–Kolmogorov Equation (2 citations)
and 7 more in the interactive map.
Bayesian optimization · 12 papers
- Parallel Bayesian Optimization of Multiple Noisy Objectives with Expected Hypervolume Improvement (40 citations)
- Improving black-box optimization in VAE latent space using decoder uncertainty (24 citations)
- Explaining Hyperparameter Optimization via Partial Dependence Plots (10 citations)
- Bayesian Optimization with High-Dimensional Outputs (9 citations)
- Combining Latent Space and Structured Kernels for Bayesian Optimization over Combinatorial Spaces (5 citations)
- Multi-Step Budgeted Bayesian Optimization with Unknown Evaluation Costs (5 citations)
- Constrained Two-step Look-Ahead Bayesian Optimization (4 citations)
- Risk-averse Heteroscedastic Bayesian Optimization (3 citations)
- Batch Multi-Fidelity Bayesian Optimization with Deep Auto-Regressive Networks (2 citations)
- Reinforced Few-Shot Acquisition Function Learning for Bayesian Optimization (2 citations)
- Bayesian Optimization of Function Networks (1 citation)
- Optimizing Conditional Value-At-Risk of Black-Box Functions
