Research map / Architectures and efficiency / RNNs and normalization
Two Sides of The Same Coin: Bridging Deep Equilibrium Models and Neural ODEs via Homotopy Continuation
Shutong Ding, Tianyu Cui, Jingya Wang, Ye Shi
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 changed | Where | Measured | Effect |
|---|---|---|---|
| Add learnable initialization (ablation) | HomoODE · CIFAR-10 | inference time: about $2.5\times$ speedup | Improved |
| Add learnable initialization (ablation) | HomoODE · CIFAR-10 | test accuracy | No clear effect |
| HomoODE vs DEQ (larger models, augmentation) | HomoODE · CIFAR-100 | accuracy: 64.5 \pm 0.7 \% -> 71.57 \pm 0.2 | Improved |
| HomoODE vs DEQ | HomoODE · Tiny-ImageNet | accuracy: 47.6 \pm 0.4 \% -> 56.1 \pm 0.1 | Improved |
| HomoODE vs Neural ODE | HomoODE · CIFAR-100 | accuracy: 31.7 \pm 0.6\% -> 71.57 \pm 0.2 | Improved |
| Train HomoODE without adjoint method instead of with adjoint | HomoODE · CIFAR-100 | accuracy: 69.30 \pm 0.1\% -> 71.57 \pm 0.2 | Improved |
| Train HomoODE without adjoint method instead of with adjoint | HomoODE · Tiny-ImageNet | accuracy: 54.14 \pm 0.2\% -> 56.1 \pm 0.1 | Improved |
| Apply data augmentation | HomoODE · CIFAR-10 | accuracy | Improved |
| Input-based initial point predictor instead of learnable initial point (ablation) | HomoODE · CIFAR-10 | inference time and accuracy | Hurt |
Closest papers by meaning
- Subhomogeneous Deep Equilibrium Models (ICML 2024 · similarity 0.88)
- Semi-Implicit Neural Ordinary Differential Equations (AAAI 2025 · similarity 0.87)
- DeltaDEQ: Exploiting Heterogeneous Convergence for Accelerating Deep Equilibrium Iterations (NeurIPS 2024 · similarity 0.87)
- Positive Concave Deep Equilibrium Models (ICML 2024 · similarity 0.87)
- Separation and Bias of Deep Equilibrium Models on Expressivity and Learning Dynamics (NeurIPS 2024 · similarity 0.87)
- How Deep Do We Need: Accelerating Training and Inference of Neural ODEs via Control Perspective (ICML 2024 · similarity 0.86)
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.
