Research map / Label-efficient and robust learning / Recognition and long-tail
Deep Combinatorial Aggregation
Yuesong Shen, Daniel Cremers
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 |
|---|---|---|---|
| train DCWA with consistency enforcing loss instead of NLL (ablation) | DCWA · CIFAR-10 | NLL: 0.2651 -> 0.2355 | Improved |
| train DCWA with consistency enforcing loss instead of NLL (ablation) | DCWA · CIFAR-10 | ECE: 0.0365 -> 0.0301 | Improved |
| train modelwise DCA with consistency enforcing loss instead of NLL (ablation) | Modelwise DCA · CIFAR-10 | NLL: 0.1601 -> 0.1550 | Improved |
| train modelwise DCA with consistency enforcing loss instead of NLL (ablation) | Modelwise DCA · CIFAR-10 | accuracy: 0.9495 -> 0.9481 | Hurt |
| train trunkwise DCA with consistency enforcing loss instead of NLL (ablation) | Trunkwise DCA · CIFAR-10 | NLL: 0.1676 -> 0.1728 | Hurt |
| use SWA-like weight averaging DCWA vs SWA | DCWA · CIFAR-10 | accuracy: 0.9327 -> 0.9337 | No clear effect |
| DCWA vs standard training | DCWA · CIFAR-10 | accuracy: 0.9264 -> 0.9337 | Improved |
| DCWA vs standard training | DCWA · SVHN | NLL: 0.1786 -> 0.1713 | Improved |
| modelwise DCA vs deep ensemble | Modelwise DCA · CIFAR-10 | NLL: 0.1746 -> 0.1550 | Improved |
| trunkwise DCA vs deep ensemble | Trunkwise DCA · CIFAR-10 | NLL: 0.1746 -> 0.1676 | Improved |
| trunkwise DCA vs deep ensemble on CIFAR-10-C | Trunkwise DCA · CIFAR-10-C | Brier and ECE | Hurt |
| coarser component granularity in DCA (ablation) | DCA · CIFAR-10 | accuracy, NLL, ECE | Improved |
Closest papers by meaning
- Hyperparameter Ensembles for Robustness and Uncertainty Quantification (NeurIPS 2020 · similarity 0.85)
- Neural Ensemble Search for Uncertainty Estimation and Dataset Shift (NeurIPS 2021 · similarity 0.85)
- Pitfalls of In-Domain Uncertainty Estimation and Ensembling in Deep Learning (ICLR 2020 · similarity 0.85)
- Combinatorial Inference against Label Noise (NeurIPS 2019 · similarity 0.85)
- Packed Ensembles for efficient uncertainty estimation (ICLR 2023 · similarity 0.85)
- SURE: SUrvey REcipes for building reliable and robust deep networks (CVPR 2024 · 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.
