Research map / Label-efficient and robust learning / Test-time adaptation
DiverseNet: When One Right Answer Is Not Enough
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 |
|---|---|---|---|
| Use DiverseNet instead of the Lee et al. ensemble | traffic-flow prediction · Traffic flow test | average endpoint error (k=4): 0.158 -> 0.146 | Helped |
| Use DiverseNet instead of L2 with test-time Dropout | traffic-flow prediction · Traffic flow test | average endpoint error (k=4): 0.184 -> 0.146 | Helped |
| Use DiverseNet instead of Treenet | ShapeNet volume prediction · ShapeNet | voxel-grid IoU (k=6): 0.545 -> 0.713 | Helped |
| Use DiverseNet instead of Treenet | ShapeNet volume prediction · ShapeNet | silhouette reprojection IoU: 0.313 -> 0.773 | Helped |
| Add Lcatchup to Treenet (ablation) | Treenet · ShapeNet | voxel-grid IoU (k=6): 0.545 -> 0.694 | Helped |
| Pretrain Treenet (ablation) | Treenet · ShapeNet | voxel-grid IoU (k=6): 0.545 -> 0.607 | Helped |
| Add Lcatchup and pretraining to Treenet (ablation) | Treenet · ShapeNet | voxel-grid IoU (k=6): 0.545 -> 0.704 | Helped |
| Add Lcatchup (ablation) | DiverseNet · ShapeNet | voxel-grid IoU (k=6): 0.704 -> 0.713 | Helped |
| Add pretraining (ablation) | DiverseNet · ShapeNet | voxel-grid IoU (k=6): 0.713 -> 0.704 | No clear effect |
| Add both pretraining and Lcatchup (ablation) | DiverseNet · ShapeNet | voxel-grid IoU (k=6): 0.713 -> 0.693 | Hurt |
| Increase the number of predictions beyond three | DiverseNet · MNISTOCC | best MSE over predictions | Helped |
| Increase beta (ablation) | DiverseNet · MNIST | best MSE over predictions | Mixed |
Closest papers by meaning
- Structured Convolutions for Efficient Neural Network Design (NeurIPS 2020 · similarity 0.85)
- Multi-Scale Dense Networks for Resource Efficient Image Classification (ICLR 2018 · similarity 0.85)
- Searching for Efficient Multi-Scale Architectures for Dense Image Prediction (NeurIPS 2018 · similarity 0.84)
- Synergy and Diversity in CLIP: Enhancing Performance Through Adaptive Backbone Ensembling (ICLR 2025 · similarity 0.84)
- Diversity Matters: Revisiting Test-Time Compute in Vision-Language Models (ICML 2026 · similarity 0.84)
- OSSCAR: One-Shot Structured Pruning in Vision and Language Models with Combinatorial Optimization (ICML 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.