Neural Ordinary Differential Equations

We introduce a new family of deep neural network models. Instead of specifying a discrete sequence of hidden layers, we parameterize the derivative of the hidden state using a neural network... (read more)

PDF Abstract NeurIPS 2018 PDF NeurIPS 2018 Abstract

Datasets


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Multivariate Time Series Forecasting MIMIC-III NeuralODE-VAE-Mask MSE 0.89 # 4
NegLL 1.36 # 5
Multivariate Time Series Forecasting MIMIC-III NeuralODE-VAE MSE 0.89 # 4
NegLL 1.35 # 4
Multivariate Time Series Imputation MuJoCo Latent ODE (RNN enc.) MSE (10^2, 50% missing) 0.447 # 2
Multivariate Time Series Imputation MuJoCo RNN-VAE MSE (10^2, 50% missing) 6.100 # 6
Multivariate Time Series Forecasting MuJoCo RNN-VAE MSE (10^-2, 50% missing) 1.782 # 3
Multivariate Time Series Forecasting MuJoCo Latent ODE (RNN enc.) MSE (10^-2, 50% missing) 1.377 # 2
Multivariate Time Series Imputation PhysioNet Challenge 2012 RNN-VAE mse (10^-3) 5.930 # 4
Multivariate Time Series Forecasting PhysioNet Challenge 2012 RNN-VAE mse (10^-3) 3.055 # 3
MSE stdev 0.145 # 4
Multivariate Time Series Forecasting PhysioNet Challenge 2012 Latent ODE (RNN enc.) mse (10^-3) 3.162 # 4
MSE stdev 0.052 # 3
Multivariate Time Series Imputation PhysioNet Challenge 2012 Latent ODE (RNN enc.) mse (10^-3) 3.907 # 3
Multivariate Time Series Forecasting USHCN-Daily NeuralODE-VAE-Mask MSE 0.83 # 4
Multivariate Time Series Forecasting USHCN-Daily NeuralODE-VAE MSE 0.96 # 6

Methods used in the Paper


METHOD TYPE
🤖 No Methods Found Help the community by adding them if they're not listed; e.g. Deep Residual Learning for Image Recognition uses ResNet