Research map / Architectures and efficiency / Neural architecture search
Winning Solutions and Post-Challenge Analyses of the ChaLearn AutoDL Challenge 2019
Zhengying Liu, Adrien Pavão, Zhen Xu, Sérgio Escalera, Fábio Ferreira, Isabelle Guyon, Sirui Hong, Frank Hutter, Rongrong Ji, Julio C. S. Jacques, Ge Li, Marius Lindauer and 13 more
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 |
|---|---|---|---|
| add meta-learning (pretrained models, tuned hyperparameters) to DeepWisdom (ablation) | DeepWisdom · AutoDL final phase (10 tasks) | ALC | Improved |
| add progressive data loading (NumPy conversion, caching) to DeepWisdom (ablation) | DeepWisdom · AutoDL final phase (10 tasks) | ALC | Improved |
| add data augmentation (fast autoaugment, TTA, stagewise spec_len) to DeepWisdom (ablation) | DeepWisdom · AutoDL final phase (image/video/speech) | ALC | Improved |
| add adaptive strategies to DeepBlueAI (ablation) | DeepBlueAI · AutoDL final phase (10 tasks) | ALC | Improved |
| add ensemble learning to DeepBlueAI (ablation) | DeepBlueAI · AutoDL final phase (10 tasks) | ALC | Improved |
| add scoring time reduction to DeepBlueAI (ablation) | DeepBlueAI · AutoDL final phase (10 tasks) | ALC | Improved |
| add AutoFolio meta-selection over complementary configurations vs random selector or generalist (ablation) | automl_freiburg · AutoDL final phase (4 image/video tasks) | ALC | Improved |
| tune AutoFolio hyperparameters with HPO on eight meta-train datasets (ablation) | automl_freiburg · AutoDL final phase (4 image/video tasks) | ALC | Hurt |
| add DeepBlueAI ensembling (EN) to automl_freiburg (ablation) | automl_freiburg · 6 image/video tasks | tasks with better ALC: 4 of 6 tasks | Improved |
| add DeepWisdom data loading (DL) to automl_freiburg (ablation) | automl_freiburg · 6 image/video tasks | tasks with better ALC: 1 of 6 tasks | No clear effect |
| add other teams' modules to DeepWisdom (ablation) | DeepWisdom · 6 image/video tasks | tasks with better ALC: 1 of 6 tasks for +EN, +HPO, +EN+HPO | Hurt |
| add other teams' modules (DL, HPO) to DeepBlueAI (ablation) | DeepBlueAI · 6 image/video tasks | tasks with better ALC: 0 of 6 tasks | Hurt |
Closest papers by meaning
- Auto-Pytorch: Multi-Fidelity MetaLearning for Efficient and Robust AutoDL (TPAMI 2021 · similarity 0.86)
- Zero-shot AutoML with Pretrained Models (ICML 2022 · similarity 0.86)
- Auto-Sklearn 2.0: Hands-free AutoML via Meta-Learning (JMLR 2022 · similarity 0.85)
- Insights from the Use of Previously Unseen Neural Architecture Search Datasets (CVPR 2024 · similarity 0.84)
- AutoAL: Automated Active Learning with Differentiable Query Strategy Search (ICML 2025 · similarity 0.84)
- TabReD: Analyzing Pitfalls and Filling the Gaps in Tabular Deep Learning Benchmarks (ICLR 2025 · similarity 0.82)
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.
