ICLR 21 paperlist

ICLR 2021 paper list

Time Series

• Multi-Time Attention Networks for Irregularly Sampled Time Series

• Discrete Graph Structure Learning for Forecasting Multiple Time Series

• Unsupervised Representation Learning for Time Series with Temporal Neighborhood Coding

• Greedy-GQ with Variance Reduction: Finite-time Analysis and Improved Complexity

• Coupled Oscillatory Recurrent Neural Network (coRNN): An accurate and (gradient) stable architecture for learning long time dependencies

Dynamic systems

• ResNet After All: Neural ODEs and Their Numerical Solution

• MALI: A memory efficient and reverse accurate integrator for Neural ODEs

• Go with the flow: Adaptive control for Neural ODEs

• Conformation-Guided Molecular Representation with Hamiltonian Neural Networks

• Identifying Physical Law of Hamiltonian Systems via Meta-Learning

• Learning continuous-time PDEs from sparse data with graph neural networks

• PDE-Driven Spatiotemporal Disentanglement

• Universal approximation power of deep residual neural networks via nonlinear control theory

• Variational Intrinsic Control Revisited

• Physics-aware, probabilistic model order reduction with guaranteed stability

• NOVAS: Non-convex Optimization via Adaptive Stochastic Search for End-to-end Learning and Control

• Score-Based Generative Modeling through Stochastic Differential Equations

• Randomized Automatic Differentiation

• Dynamics of Deep Equilibrium Linear Models

• Identifying nonlinear dynamical systems with multiple time scales and long-range dependencies

Generative Models

• Projected Latent Markov Chain Monte Carlo: Conditional Sampling of Normalizing Flows

• Categorical Normalizing Flows via Continuous Transformations

• Generative Time-series Modeling with Fourier Flows

• IDF++: Analyzing and Improving Integer Discrete Flows for Lossless Compression

• Refining Deep Generative Models via Wasserstein Gradient Flows

• Convex Potential Flows: Universal Probability Distributions with Optimal Transport and Convex Optimization

• Flowtron: an Autoregressive Flow-based Generative Network for Text-to-Speech Synthesis

• Decoupling Global and Local Representations via Invertible Generative Flows

• Combining Physics and Machine Learning for Network Flow Estimation

• Enjoy Your Editing: Controllable GANs for Image Editing via Latent Space Navigation

• Property Controllable Variational Autoencoder via Invertible Mutual Dependence