Research map / Label-efficient and robust learning / Recognition and long-tail
Deep Discriminative CNN with Temporal Ensembling for Ambiguously-Labeled Image Classification
Yao Yao, Jiehui Deng, Xiuhua Chen, Chen Gong, Jianxin Wu, Jian Yang
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 |
|---|---|---|---|
| replace cross-entropy DCNN with D2CNN loss (entropy + temporal ensembling) | ResNet-20 · Lost | accuracy: 0.580 ± 0.031 -> 0.838 ± 0.014 | Improved |
| replace cross-entropy DCNN with D2CNN loss | ResNet-20 · Yahoo!News | accuracy: 0.740 ± 0.006 -> 0.833 ± 0.009 | Improved |
| replace cross-entropy DCNN with D2CNN loss | ResNet-20 · SVHN-v3 | accuracy: 0.908 ± 0.007 -> 0.929 ± 0.001 | Improved |
| add discrimination (entropy) term Ld (ablation) | D2CNN · Lost | accuracy | Improved |
| add temporal ensembling term Lt (ablation) | D2CNN · Lost | accuracy | Improved |
| ramp weight T(t) up from zero instead of fixed T(t)=Tmax (ablation) | D2CNN · Lost | accuracy | Improved |
| feed deep features instead of GIST features to baselines | non-deep baselines (e.g. IPAL) · Lost | accuracy: 0.790 ± 0.034 -> 0.824 ± 0.021 | Improved |
| use D2CNN instead of SURE with deep features | deep-feature baselines · Lost | accuracy: 0.826 ± 0.006 -> 0.838 ± 0.014 | Improved |
| increase false-positive candidate labels from 1 to 3 | D2CNN · SVHN | accuracy: 0.937 ± 0.003 -> 0.929 ± 0.001 | No clear effect |
Closest papers by meaning
- Multilabel Deep Visual-Semantic Embedding (TPAMI 2019 · similarity 0.88)
- CNN-RNN: A Unified Framework for Multi-Label Image Classification (CVPR 2016 · similarity 0.88)
- Learning Spatial Regularization With Image-Level Supervisions for Multi-Label Image Classification (CVPR 2017 · similarity 0.87)
- Collaborative Layer-Wise Discriminative Learning in Deep Neural Networks (ECCV 2016 · similarity 0.87)
- HCP: A Flexible CNN Framework for Multi-Label Image Classification (TPAMI 2016 · similarity 0.87)
- Multi-Label Image Recognition by Recurrently Discovering Attentional Regions (ICCV 2017 · 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.
