Research map / Architectures and efficiency / CNN design
Refining Architectures of Deep Convolutional Neural Networks
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Apply the combined stretch-and-split refinement | VGG-11 and GoogleNet · SUN Attributes Dataset (42 attributes) | precision@21 | Helped |
| Apply the combined stretch-and-split refinement | VGG-11 and GoogleNet · CAMIT-NSAD | precision@7 | Hurt |
| Add symmetric splitting to stretch-only refinement (ablation) | DR-1 (stretch-only refinement) on VGG-11 and GoogleNet · SUN Attributes Dataset (42 attributes) | precision@21 | Helped |
| Add stretching to split-only refinement (ablation) | DR-2 (split-only refinement) on VGG-11 and GoogleNet · SUN Attributes Dataset (42 attributes) | precision@21 | Helped |
| Use the combined stretch-and-split refinement instead of sparsification | VGG-11 and GoogleNet · SUN Attributes Dataset (42 attributes) | precision@21 | Helped |
| Use the refined architecture instead of sparsification | DR and DR-2 · CAMIT-NSAD | precision@7 | Helped |
| Increase λ (ablation) | Architecture refinement on CAMIT-NSAD · CAMIT-NSAD | model-size reduction | Hurt |
| Randomly stretch and split throughout the network instead of selecting operations using class-separation statistics | CNN architecture refinement · SUN Attributes Dataset and CAMIT-NSAD | precision@k | Hurt |
Closest papers by meaning
- An Empirical Evaluation of Current Convolutional Architectures' Ability to Manage Nuisance Location and Scale Variability (CVPR 2016 · similarity 0.89)
- EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks (ICML 2019 · similarity 0.88)
- Deep Pyramidal Residual Networks (CVPR 2017 · similarity 0.88)
- Structured Convolutions for Efficient Neural Network Design (NeurIPS 2020 · similarity 0.88)
- The Unreasonable Effectiveness of Fully-Connected Layers for Low-Data Regimes (NeurIPS 2022 · similarity 0.88)
- Principled Architecture-aware Scaling of Hyperparameters (ICLR 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.