ICML 2021 accepted papers on Probabilistic and scientific modeling
147 of the 1,184 papers accepted at ICML 2021 (12.4%) are on Probabilistic and scientific modeling, compared with 15.7% at ICML 2020. They fall into 8 topics.
See ICML 2021 in the interactive map
Submitting to ICML? 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 · 58 papers
- Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech (121 citations)
- Simple and Effective VAE Training with Calibrated Decoders (33 citations)
- Understanding Failures in Out-of-Distribution Detection with Deep Generative Models (22 citations)
- An Identifiable Double VAE For Disentangled Representations (11 citations)
- Differentiable Particle Filtering via Entropy-Regularized Optimal Transport (11 citations)
- Finite mixture models do not reliably learn the number of components (11 citations)
- Monte Carlo Variational Auto-Encoders (10 citations)
- Sliced Iterative Normalizing Flows (9 citations)
- Hierarchical VAEs Know What They Don’t Know (8 citations)
- Oops I Took A Gradient: Scalable Sampling for Discrete Distributions (8 citations)
- Deep Generative Learning via Schrödinger Bridge (5 citations)
- Demystifying Inductive Biases for (Beta-)VAE Based Architectures (5 citations)
and 46 more in the interactive map.
Uncertainty and Gaussian processes · 28 papers
- What Are Bayesian Neural Network Posteriors Really Like? (99 citations)
- PACOH: Bayes-Optimal Meta-Learning with PAC-Guarantees (38 citations)
- Classifying high-dimensional Gaussian mixtures: Where kernel methods fail and neural networks succeed (28 citations)
- Global inducing point variational posteriors for Bayesian neural networks and deep Gaussian processes (17 citations)
- GP-Tree: A Gaussian Process Classifier for Few-Shot Incremental Learning (14 citations)
- Scalable Marginal Likelihood Estimation for Model Selection in Deep Learning (11 citations)
- Bayesian Deep Learning via Subnetwork Inference (10 citations)
- Deep Adaptive Design: Amortizing Sequential Bayesian Experimental Design (8 citations)
- Amortized Conditional Normalized Maximum Likelihood: Reliable Out of Distribution Uncertainty Estimation (6 citations)
- Nondeterminism and Instability in Neural Network Optimization (6 citations)
- The Hintons in your Neural Network: a Quantum Field Theory View of Deep Learning (6 citations)
- A Bit More Bayesian: Domain-Invariant Learning with Uncertainty (5 citations)
and 16 more in the interactive map.
Molecules and proteins · 27 papers
- Learning Gradient Fields for Molecular Conformation Generation (39 citations)
- Neural Symbolic Regression that scales (26 citations)
- An End-to-End Framework for Molecular Conformation Generation via Bilevel Programming (20 citations)
- Fold2Seq: A Joint Sequence(1D)-Fold(3D) Embedding-based Generative Model for Protein Design (16 citations)
- Non-Autoregressive Electron Redistribution Modeling for Reaction Prediction (11 citations)
- Diffusion Earth Mover’s Distance and Distribution Embeddings (10 citations)
- MorphVAE: Generating Neural Morphologies from 3D-Walks using a Variational Autoencoder with Spherical Latent Space (9 citations)
- Latent Programmer: Discrete Latent Codes for Program Synthesis (7 citations)
- A Structured Observation Distribution for Generative Biological Sequence Prediction and Forecasting (6 citations)
- Benchmarks, Algorithms, and Metrics for Hierarchical Disentanglement (6 citations)
- Commutative Lie Group VAE for Disentanglement Learning (6 citations)
- Event Outlier Detection in Continuous Time (4 citations)
and 15 more in the interactive map.
Causal discovery · 8 papers
- Necessary and sufficient conditions for causal feature selection in time series with latent common causes (12 citations)
- Causality-aware counterfactual confounding adjustment as an alternative to linear residualization in anticausal prediction tasks based on linear learners (6 citations)
- Integer Programming for Causal Structure Learning in the Presence of Latent Variables (5 citations)
- Estimating Identifiable Causal Effects on Markov Equivalence Class through Double Machine Learning (2 citations)
- A Proxy Variable View of Shared Confounding (1 citation)
- Cumulants of Hawkes Processes are Robust to Observation Noise (1 citation)
- Operationalizing Complex Causes: A Pragmatic View of Mediation (1 citation)
- Regularizing towards Causal Invariance: Linear Models with Proxies (1 citation)
Dynamical systems · 7 papers
- Inferring Latent Dynamics Underlying Neural Population Activity via Neural Differential Equations (20 citations)
- Data-driven Prediction of General Hamiltonian Dynamics via Learning Exactly-Symplectic Maps (19 citations)
- Variational Data Assimilation with a Learned Inverse Observation Operator (19 citations)
- Temporal Predictive Coding For Model-Based Planning In Latent Space (5 citations)
- 12-Lead ECG Reconstruction via Koopman Operators (2 citations)
- Prediction-Centric Learning of Independent Cascade Dynamics from Partial Observations (1 citation)
- Segmenting Hybrid Trajectories using Latent ODEs
Treatment effects · 7 papers
- Quantifying Ignorance in Individual-Level Causal-Effect Estimates under Hidden Confounding (12 citations)
- Proximal Causal Learning with Kernels: Two-Stage Estimation and Moment Restriction (9 citations)
- Valid Causal Inference with (Some) Invalid Instruments (6 citations)
- Conditional Distributional Treatment Effect with Kernel Conditional Mean Embeddings and U-Statistic Regression (5 citations)
- Budgeted Heterogeneous Treatment Effect Estimation (4 citations)
- How and Why to Use Experimental Data to Evaluate Methods for Observational Causal Inference (3 citations)
- Permutation Weighting (3 citations)
Differential equations and PDEs · 6 papers
- Solving high-dimensional parabolic PDEs using the tensor train format (8 citations)
- Neural SDEs as Infinite-Dimensional GANs (7 citations)
- Learning Stochastic Behaviour from Aggregate Data (6 citations)
- Opening the Blackbox: Accelerating Neural Differential Equations by Regularizing Internal Solver Heuristics (4 citations)
- "Hey, that’s not an ODE": Faster ODE Adjoints via Seminorms (2 citations)
- Functional Space Analysis of Local GAN Convergence (2 citations)
Bayesian optimization · 6 papers
- Think Global and Act Local: Bayesian Optimisation over High-Dimensional Categorical and Mixed Search Spaces (14 citations)
- Bayesian Algorithm Execution: Estimating Computable Properties of Black-box Functions Using Mutual Information (4 citations)
- Bayesian Optimization over Hybrid Spaces (4 citations)
- BORE: Bayesian Optimization by Density-Ratio Estimation (2 citations)
- Bayesian Optimistic Optimisation with Exponentially Decaying Regret
- Objective Bound Conditional Gaussian Process for Bayesian Optimization
