ICLR 2022 accepted papers on Probabilistic and scientific modeling
109 of the 1,095 papers accepted at ICLR 2022 (10.0%) are on Probabilistic and scientific modeling, compared with 10.2% at ICLR 2021. They fall into 8 topics.
See ICLR 2022 in the interactive map
Submitting to ICLR? 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 · 37 papers
- How to deal with missing data in supervised deep learning? (17 citations)
- Variational methods for simulation-based inference (10 citations)
- A Deep Variational Approach to Clustering Survival Data (8 citations)
- A Tale of Two Flows: Cooperative Learning of Langevin Flow and Normalizing Flow Toward Energy-Based Model (8 citations)
- Exposing the Implicit Energy Networks behind Masked Language Models via Metropolis--Hastings (7 citations)
- Likelihood Training of Schrödinger Bridge using Forward-Backward SDEs Theory (6 citations)
- Variational Neural Cellular Automata (6 citations)
- Autoregressive Quantile Flows for Predictive Uncertainty Estimation (4 citations)
- Differentiable DAG Sampling (4 citations)
- Knowledge Removal in Sampling-based Bayesian Inference (4 citations)
- MCMC Should Mix: Learning Energy-Based Model with Neural Transport Latent Space MCMC (3 citations)
- Energy-Based Learning for Cooperative Games, with Applications to Valuation Problems in Machine Learning (2 citations)
and 25 more in the interactive map.
Molecules and proteins · 28 papers
- GeoDiff: A Geometric Diffusion Model for Molecular Conformation Generation (179 citations)
- Crystal Diffusion Variational Autoencoder for Periodic Material Generation (105 citations)
- Iterative Refinement Graph Neural Network for Antibody Sequence-Structure Co-design (66 citations)
- OntoProtein: Protein Pretraining With Gene Ontology Embedding (49 citations)
- Learning to Extend Molecular Scaffolds with Structural Motifs (36 citations)
- Independent SE(3)-Equivariant Models for End-to-End Rigid Protein Docking (26 citations)
- Amortized Tree Generation for Bottom-up Synthesis Planning and Synthesizable Molecular Design (20 citations)
- Geometric Transformers for Protein Interface Contact Prediction (15 citations)
- MIDI-DDSP: Detailed Control of Musical Performance via Hierarchical Modeling (10 citations)
- Do Not Escape From the Manifold: Discovering the Local Coordinates on the Latent Space of GANs (8 citations)
- GeneDisco: A Benchmark for Experimental Design in Drug Discovery (8 citations)
- Ab-Initio Potential Energy Surfaces by Pairing GNNs with Neural Wave Functions (4 citations)
and 16 more in the interactive map.
Dynamical systems · 17 papers
- iLQR-VAE : control-based learning of input-driven dynamics with applications to neural data (30 citations)
- Deconstructing the Inductive Biases of Hamiltonian Neural Networks (13 citations)
- Learning the Dynamics of Physical Systems from Sparse Observations with Finite Element Networks (10 citations)
- Constrained Physical-Statistics Models for Dynamical System Identification and Prediction (6 citations)
- Fine-grained Differentiable Physics: A Yarn-level Model for Fabrics (5 citations)
- The Three Stages of Learning Dynamics in High-dimensional Kernel Methods (2 citations)
- Filtered-CoPhy: Unsupervised Learning of Counterfactual Physics in Pixel Space (1 citation)
- Hidden Parameter Recurrent State Space Models For Changing Dynamics Scenarios (1 citation)
- Variational Predictive Routing with Nested Subjective Timescales
- Connectome-constrained Latent Variable Model of Whole-Brain Neural Activity
- Curriculum learning as a tool to uncover learning principles in the brain
- D-CODE: Discovering Closed-form ODEs from Observed Trajectories
and 5 more in the interactive map.
Uncertainty and Gaussian processes · 9 papers
- On the Pitfalls of Heteroscedastic Uncertainty Estimation with Probabilistic Neural Networks (41 citations)
- Transformers Can Do Bayesian Inference (25 citations)
- Bayesian Neural Network Priors Revisited (23 citations)
- Natural Posterior Network: Deep Bayesian Predictive Uncertainty for Exponential Family Distributions (11 citations)
- Practical Conditional Neural Process Via Tractable Dependent Predictions (5 citations)
- Counterfactual Plans under Distributional Ambiguity (4 citations)
- Scale Mixtures of Neural Network Gaussian Processes (2 citations)
- Bayesian Modeling and Uncertainty Quantification for Learning to Optimize: What, Why, and How
- Neural Variational Dropout Processes
Differential equations and PDEs · 8 papers
- Message Passing Neural PDE Solvers (58 citations)
- Unsupervised Learning of Full-Waveform Inversion: Connecting CNN and Partial Differential Equation in a Loop (24 citations)
- Discovering Nonlinear PDEs from Scarce Data with Physics-encoded Learning (12 citations)
- Steerable Partial Differential Operators for Equivariant Neural Networks (5 citations)
- Neural Solvers for Fast and Accurate Numerical Optimal Control (1 citation)
- Blaschke Product Neural Networks (BPNN): A Physics-Infused Neural Network for Phase Retrieval of Meromorphic Functions
- Non-Linear Operator Approximations for Initial Value Problems
- SketchODE: Learning neural sketch representation in continuous time
Causal discovery · 7 papers
- Learning Temporally Causal Latent Processes from General Temporal Data (6 citations)
- Granger causal inference on DAGs identifies genomic loci regulating transcription (5 citations)
- Efficient Neural Causal Discovery without Acyclicity Constraints (2 citations)
- Optimal Transport for Causal Discovery (1 citation)
- Learning Causal Models from Conditional Moment Restrictions by Importance Weighting
- Properties from mechanisms: an equivariance perspective on identifiable representation learning
- Invariant Causal Representation Learning for Out-of-Distribution Generalization
