ICLR 2020 accepted papers on Probabilistic and scientific modeling
59 of the 687 papers accepted at ICLR 2020 (8.6%) are on Probabilistic and scientific modeling, compared with 12.2% at ICLR 2019. They fall into 8 topics.
See ICLR 2020 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 · 21 papers
- Input Complexity and Out-of-distribution Detection with Likelihood-based Generative Models (138 citations)
- GENESIS: Generative Scene Inference and Sampling with Object-Centric Latent Representations (130 citations)
- From Variational to Deterministic Autoencoders (103 citations)
- Diverse Trajectory Forecasting with Determinantal Point Processes (90 citations)
- Empirical Bayes Transductive Meta-Learning with Synthetic Gradients (81 citations)
- A Probabilistic Formulation of Unsupervised Text Style Transfer (63 citations)
- Disentanglement by Nonlinear ICA with General Incompressible-flow Networks (GIN) (46 citations)
- Understanding the Limitations of Variational Mutual Information Estimators (29 citations)
- The Shape of Data: Intrinsic Distance for Data Distributions (24 citations)
- Estimating Gradients for Discrete Random Variables by Sampling without Replacement (14 citations)
- Unbiased Contrastive Divergence Algorithm for Training Energy-Based Latent Variable Models (14 citations)
- Mixed-curvature Variational Autoencoders (12 citations)
and 9 more in the interactive map.
Molecules and proteins · 13 papers
- Sequential Latent Knowledge Selection for Knowledge-Grounded Dialogue (121 citations)
- Data-dependent Gaussian Prior Objective for Language Generation (53 citations)
- RNA Secondary Structure Prediction By Learning Unrolled Algorithms (36 citations)
- A Theory of Usable Information under Computational Constraints (30 citations)
- Counterfactuals uncover the modular structure of deep generative models (28 citations)
- Mathematical Reasoning in Latent Space (24 citations)
- Convolutional Conditional Neural Processes (13 citations)
- Energy-based models for atomic-resolution protein conformations (13 citations)
- A Latent Morphology Model for Open-Vocabulary Neural Machine Translation (8 citations)
- Dynamic Time Lag Regression: Predicting What & When (6 citations)
- Decoding As Dynamic Programming For Recurrent Autoregressive Models (4 citations)
- PCMC-Net: Feature-based Pairwise Choice Markov Chains (1 citation)
and 1 more in the interactive map.
Dynamical systems · 10 papers
- Symplectic ODE-Net: Learning Hamiltonian Dynamics with Control (128 citations)
- Symplectic Recurrent Neural Networks (103 citations)
- Lagrangian Fluid Simulation with Continuous Convolutions (76 citations)
- Hamiltonian Generative Networks (52 citations)
- Learning Compositional Koopman Operators for Model-Based Control (50 citations)
- Physics-aware Difference Graph Networks for Sparsely-Observed Dynamics (36 citations)
- CoPhy: Counterfactual Learning of Physical Dynamics (34 citations)
- Intrinsically Motivated Discovery of Diverse Patterns in Self-Organizing Systems (15 citations)
- Prediction, Consistency, Curvature: Representation Learning for Locally-Linear Control (8 citations)
- Implementing Inductive bias for different navigation tasks through diverse RNN attrractors
Uncertainty and Gaussian processes · 6 papers
- Learning to Balance: Bayesian Meta-Learning for Imbalanced and Out-of-distribution Tasks (19 citations)
- Conservative Uncertainty Estimation By Fitting Prior Networks (18 citations)
- Quantifying Point-Prediction Uncertainty in Neural Networks via Residual Estimation with an I/O Kernel (12 citations)
- Hypermodels for Exploration (10 citations)
- Bayesian Meta Sampling for Fast Uncertainty Adaptation (8 citations)
- Sampling-Free Learning of Bayesian Quantized Neural Networks (3 citations)
Differential equations and PDEs · 3 papers
- Deep Learning For Symbolic Mathematics (46 citations)
- Learning to Control PDEs with Differentiable Physics (37 citations)
- SNODE: Spectral Discretization of Neural ODEs for System Identification (20 citations)
