AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Symplectic Adjoint Method for Exact Gradient of Neural ODE with Minimal Memory

Takashi Matsubara, Yuto Miyatake, Takaharu Yaguchi

NeurIPS 2021 · Architectures and efficiency › RNNs and normalization · approach: Scalable approximation and equivariant compute · 5 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

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
Use the symplectic adjoint method instead of the adjoint methodNeural ODE training · MiniBooNEpeak memory (MiB): 170 -> 95Improved
Use the symplectic adjoint method instead of the adjoint methodNeural ODE training · MiniBooNEtime (s/itr): 0.74 -> 0.84Hurt
Use the symplectic adjoint method instead of ACANeural ODE training · GASpeak memory (MiB): 73 -> 20Improved
Use the symplectic adjoint method instead of ACANeural ODE training · GAStime (s/itr): 3.98 -> 4.39Hurt
Use the symplectic adjoint method instead of ACANeural ODE training · HEPMASSpeak memory (MiB): 88 -> 35Improved
Use the symplectic adjoint method instead of ACANeural ODE training · HEPMASStime (s/itr): 3.67 -> 4.15Hurt
Use the symplectic adjoint method instead of ACANeural ODE training · BSDS300peak memory (MiB): 757 -> 283Improved
Use the symplectic adjoint method instead of ACANeural ODE training · BSDS300time (s/itr): 6.97 -> 8.07Hurt
Use the symplectic adjoint method instead of the adjoint methodNeural ODE training · MNISTtime (s/itr): 10.12 -> 9.42Improved
Use the symplectic adjoint method instead of ACANeural ODE training · MNISTpeak memory (MiB): 4,332 -> 1,079Improved
Use the symplectic adjoint method instead of ACANeural ODE training · MNISTtime (s/itr): 7.94 -> 9.42Hurt
Use the symplectic adjoint method rather than ordinary backpropagationContinuous normalizing flows · GASNLL: -9.53 -> -10.89Improved

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