AI Paper Atlas

Research map / Graph learning / Graph algorithms and clustering

Adaptive Checkpoint Adjoint Method for Gradient Estimation in Neural ODE

Juntang Zhuang, Nicha Dvornek, Xiaoxiao Li, Sekhar Tatikonda, Xenophon Papademetris, James Duncan

ICML 2020 · Graph learning › Graph algorithms and clustering · approach: Neural combinatorial optimization on graphs · 53 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Train NODE18 with ACA instead of the adjoint methodNODE18 · CIFAR10 testerror rate: 4.92 -> 4.85Improved
Train NODE18 with ACA instead of the naive methodNODE18 · CIFAR10 testerror rate: 5.29 -> 4.85Improved
Use ACA-trained NODE18 rather than ResNet18image classification · CIFAR10 testerror rate: 9.8 (*19) -> 4.85Improved
Test ACA-trained NODE18 with alternate adaptive solvers (ablation)NODE18-ACA · CIFAR10 and CIFAR100 testerror rate: CIFAR10: 4.85 -> 5.27, 5.24, 5.29; CIFAR100: 22.66 -> 24.44, 24.43, 24.43No clear effect
Train Latent-ODE with ACA instead of the adjoint methodLatent-ODE · Mujoco irregularly sampled time seriestest MSE (×10−2): 0.47 -> 0.31Improved
Train Latent-ODE with ACA instead of the naive methodLatent-ODE · Mujoco irregularly sampled time seriestest MSE (×10−2): *0.36 -> 0.31Improved
Add partial physical information through augmented inputsLSTM · Three-body future trajectoryTest MSE: 0.59+-0.12 -> 0.49+-0.06Improved
Use NODE-ACA with partial physical information rather than LSTM-aug-inputthree-body trajectory model · Three-body future trajectoryTest MSE: 0.49+-0.06 -> 0.16+-0.06Improved
Train the physical ODE model with ACA instead of the adjoint methodODE model · Three-body future trajectoryTest MSE: 0.0025+-0.0012 -> 0.0007+-0.0005Improved
Train the physical ODE model with ACA instead of the naive methodODE model · Three-body future trajectoryTest MSE: 0.0025+-0.0013 -> 0.0007+-0.0005Improved
Use ACA instead of adjoint gradient estimationNODE18 · CIFAR100 testerror rate: 24.13 -> 22.66Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy