Research map / Learning theory and algorithms / Prediction and decision losses
Self-Born Wiring for Neural Trees
Ying Chen, Feng Mao, Jie Song, Xinchao Wang, Huiqiong Wang, Mingli Song
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 |
|---|---|---|---|
| use self-born wiring instead of random wiring (ablation) | SeBoW-A · CIFAR-10 | accuracy (%): 83.32 -> 93.45 (same 1.0M params) | Improved |
| use self-born wiring instead of random wiring (ablation) | SeBoW-C · CIFAR-100 | accuracy (%): 67.04 -> 74.59 | Improved |
| use self-born wiring instead of random wiring (ablation) | SeBoW-B · CIFAR-100 | accuracy (%): 63.42 -> 71.79 | Improved |
| use self-born wiring instead of random wiring (ablation) | SeBoW-B · CIFAR-10 | accuracy (%): 84.04 -> 94.00 | Improved |
| add senders (ablation) | SeBoW-B · CIFAR-100 | accuracy (%): 52.89 -> 71.79 | Improved |
| add senders (ablation) | SeBoW-C · ImageNet | top-1 accuracy (%): 62.29 -> 70.13 | Improved |
| add senders (ablation) | SeBoW-C · tiny-ImageNet | accuracy (%): 42.01 -> 58.77 | Improved |
| add receivers (ablation) | SeBoW-A · CIFAR-10 | accuracy (%): 92.55 -> 93.45 | Improved |
| add receivers (ablation) | SeBoW-C · tiny-ImageNet | accuracy (%): 54.42 -> 58.77 | Improved |
| add receivers (ablation) | SeBoW-C · CIFAR-100 | accuracy (%): 74.59 -> 74.59 | No clear effect |
| use single-path instead of multi-path inference | SeBoW-C · ImageNet | top-1 accuracy (%): 70.13 -> 69.86; params 16.90M -> 14.78M | Hurt |
| increase base learner width from SeBoW-A to SeBoW-C | SeBoW · CIFAR-10 | accuracy (%): 93.45 -> 94.33 | Improved |
Closest papers by meaning
- Discovering Neural Wirings (NeurIPS 2019 · similarity 0.85)
- Adaptive Neural Trees (ICML 2019 · similarity 0.85)
- Differentiable Dynamic Wirings for Neural Networks (ICCV 2021 · similarity 0.84)
- Neural Trees for Learning on Graphs (NeurIPS 2021 · similarity 0.84)
- Building Variable-Sized Models via Learngene Pool (AAAI 2024 · similarity 0.82)
- Scalable Neural Network Verification with Branch-and-bound Inferred Cutting Planes (NeurIPS 2024 · similarity 0.81)
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.
