Research map / Architectures and efficiency / CNN design
Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning
Read the paperCheck 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 |
|---|---|---|---|
| Add residual connections to an Inception model at similar computational cost | Inception-v3 / Inception-ResNet-v1 · ILSVRC 2012 validation, single crop | training speed | Helped |
| Add residual connections to an Inception model at similar computational cost | Inception-v3 / Inception-ResNet-v1 · ILSVRC 2012 validation, single crop | Top-1 Error: 21.2% -> 21.3% | Hurt |
| Add residual connections to an Inception model at similar computational cost | Inception-v3 / Inception-ResNet-v1 · ILSVRC 2012 validation, single crop | Top-5 Error: 5.6% -> 5.5% | Helped |
| Use a larger, streamlined pure Inception architecture | Inception-v3 · ILSVRC 2012 validation, single crop | Top-1 Error: 21.2% -> 20.0% | Helped |
| Use a larger, streamlined pure Inception architecture | Inception-v3 · ILSVRC 2012 validation, single crop | Top-5 Error: 5.6% -> 5.0% | Helped |
| Add residual connections to the larger Inception architecture | Inception-v4 / Inception-ResNet-v2 · ILSVRC 2012 validation, single crop | Top-1 Error: 20.0% -> 19.9% | Helped |
| Add residual connections to the larger Inception architecture | Inception-v4 / Inception-ResNet-v2 · ILSVRC 2012 validation, single crop | Top-5 Error: 5.0% -> 4.9% | Helped |
| Scale residual activations before adding them to the preceding layer | Wide Inception-ResNet · ImageNet training | training stability | Helped |
| Ensemble Inception-v4 with three Inception-ResNet-v2 models | Inception-v3 ensemble · ILSVRC 2012 validation, ensemble | Top-5 Error: 3.6% -> 3.1% | Helped |
| Ensemble Inception-v4 with three Inception-ResNet-v2 models | Inception-v3 ensemble · ILSVRC 2012 validation, ensemble | Top-1 Error: 17.3% -> 16.5% | Helped |
Closest papers by meaning
- Rethinking the Inception Architecture for Computer Vision (CVPR 2016 · similarity 0.92)
- Deep Residual Learning for Image Recognition (CVPR 2016 · similarity 0.90)
- Residual Alignment: Uncovering the Mechanisms of Residual Networks (NeurIPS 2023 · similarity 0.88)
- InceptionNeXt: When Inception Meets ConvNeXt (CVPR 2024 · similarity 0.88)
- Revisiting ResNets: Improved Training and Scaling Strategies (NeurIPS 2021 · similarity 0.87)
- Revisiting Residual Connections: Orthogonal Updates for Stable and Efficient Deep Networks (NeurIPS 2025 · similarity 0.87)
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.