Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Augmented Neural ODEs

About

We show that Neural Ordinary Differential Equations (ODEs) learn representations that preserve the topology of the input space and prove that this implies the existence of functions Neural ODEs cannot represent. To address these limitations, we introduce Augmented Neural ODEs which, in addition to being more expressive models, are empirically more stable, generalize better and have a lower computational cost than Neural ODEs.

Emilien Dupont, Arnaud Doucet, Yee Whye Teh• 2019

Related benchmarks

TaskDatasetResultRank
Image ClassificationMNIST (test)
Accuracy98.2
894
ForecastingMuJoCo 70% Dropped (test)
MSE0.057
12
ForecastingMuJoCo Regular (test)
MSE0.055
12
ForecastingMuJoCo 30% Dropped (test)
MSE0.056
12
ForecastingMuJoCo 50% Dropped (test)
MSE0.057
12
ForecastingGoogle stock data 70% Dropped scenario
MSE0.0023
11
ForecastingGoogle stock data Regular scenario
MSE0.0023
11
ForecastingGoogle stock data 30% Dropped scenario
MSE0.0025
11
ForecastingGoogle stock data 50% Dropped scenario
MSE0.0029
11
Dynamics PredictionDuffing oscillator noise level 0
MSE (acceleration)3.15
7
Showing 10 of 17 rows

Other info

Code

Follow for update