Research map / Label-efficient and robust learning / Recognition and long-tail
Explore Visual Concept Formation for Image Classification
Shengzhou Xiong, Yihua Tan, Guoyou Wang
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 |
|---|---|---|---|
| replace one-hot CE with LSOVCF EE-Loss (learned Gaussian concept descriptions + experience cache) | ResNet-18 · CIFAR10 | accuracy (%): +0.50 | Improved |
| replace one-hot CE with LSOVCF EE-Loss | VGG-11-BN · CIFAR10 | accuracy (%): +0.12 | No clear effect |
| replace one-hot CE with LSOVCF EE-Loss | ShuffleNet v2 · CIFAR10 | accuracy (%): +0.47 | Improved |
| replace one-hot CE with LSOVCF EE-Loss | MobileNet v3, SENet, EfficientNet, GhostNet · CIFAR10 | accuracy (%): small gains (0.07 to 0.34) | Mixed |
| replace one-hot CE with LSOVCF EE-Loss | ResNet-18 · STL10 | accuracy (%): +2.18 | Improved |
| replace one-hot CE with LSOVCF EE-Loss | VGG-11-BN · STL10 | accuracy (%): -0.27 | Hurt |
| replace one-hot CE with LSOVCF EE-Loss | EfficientNet-b0, SENet, MobileNet v3, ShuffleNet v2, GhostNet · STL10 | accuracy (%): gains of 0.33 to 2.52 | Improved |
| replace one-hot CE with LSOVCF EE-Loss | ResNet-18 · Flower17 | accuracy (%): +5.76 | Improved |
| replace one-hot CE with LSOVCF EE-Loss | GhostNet, VGG-11-BN · Flower17 | accuracy (%): GhostNet -0.36; VGG -0.71 | Hurt |
| replace one-hot CE with LSOVCF EE-Loss | ResNet-18 · ImageNet-100 subset | accuracy (%): +1.89 | Improved |
| replace one-hot CE with LSOVCF EE-Loss | ShuffleNet v2 · ImageNet-100 subset | accuracy (%): -1.13 | Hurt |
| replace one-hot CE with LSOVCF EE-Loss | EfficientNet-b0 and GhostNet · ImageNet-100 subset | accuracy (%): EfficientNet +0.42; GhostNet +0.26 | Improved |
Closest papers by meaning
- Visual Concept Connectome (VCC): Open World Concept Discovery and their Interlayer Connections in Deep Models (CVPR 2024 · similarity 0.85)
- Language-Informed Visual Concept Learning (ICLR 2024 · similarity 0.83)
- ConceptBed: Evaluating Concept Learning Abilities of Text-to-Image Diffusion Models (AAAI 2024 · similarity 0.83)
- CLiC: Concept Learning in Context (CVPR 2024 · similarity 0.83)
- Bridging the gap to real-world language-grounded visual concept learning (NeurIPS 2025 · similarity 0.83)
- Self-Evolving Visual Concept Library using Vision-Language Critics (CVPR 2025 · similarity 0.83)
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.
