Winning solutions and post-challenge analyses of the ChaLearn AutoDL challenge 2019 - Archive ouverte HAL
Article Dans Une Revue IEEE Transactions on Pattern Analysis and Machine Intelligence Année : 2021

Winning solutions and post-challenge analyses of the ChaLearn AutoDL challenge 2019

Zhengying Liu
  • Fonction : Auteur
  • PersonId : 1038060
Zhen Xu
  • Fonction : Auteur
Rongrong Ji
  • Fonction : Auteur
Ge Li
  • Fonction : Auteur
Jin Wang
  • Fonction : Auteur
Peng Wang
Yang Zhang

Résumé

This paper reports the results and post-challenge analyses of ChaLearn’s AutoDL challenge series, which helped sorting out a profusion of AutoML solutions for Deep Learning (DL) that had been introduced in a variety of settings, but lacked fair comparisons. All input data modalities (time series, images, videos, text, tabular) were formatted as tensors and all tasks were multi-label classification problems. Code submissions were executed on hidden tasks, with limited time and computational resources, pushing solutions that get results quickly. In this setting, DL methods dominated, though popular Neural Architecture Search (NAS) was impractical. Solutions relied on fine-tuned pre-trained networks, with architectures matching data modality. Post-challenge tests did not reveal improvements beyond the imposed time limit. While no component is particularly original or novel, a high level modular organization emerged featuring a “meta-learner”, “data ingestor”, “model selector”, “model/learner”, and “evaluator”. This modularity enabled ablation studies, which revealed the importance of (off-platform) meta-learning, ensembling, and efficient data management. Experiments on heterogeneous module combinations further confirm the (local) optimality of the winning solutions. Our challenge legacy includes an ever-lasting benchmark (http://autodl.chalearn.org), the open-sourced code of the winners, and a free “AutoDL self-service”.
Fichier principal
Vignette du fichier
Winning solutions and post-challenge analyses of the ChaLearn AutoDL challenge 2019 - DRAFT.pdf (1.87 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-02957135 , version 1 (04-10-2020)
hal-02957135 , version 2 (30-11-2020)
hal-02957135 , version 3 (06-05-2021)
hal-02957135 , version 4 (01-12-2021)
hal-02957135 , version 5 (08-01-2022)

Identifiants

  • HAL Id : hal-02957135 , version 3

Citer

Zhengying Liu, Adrien Pavao, Zhen Xu, Sergio Escalera, Fabio Ferreira, et al.. Winning solutions and post-challenge analyses of the ChaLearn AutoDL challenge 2019. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021. ⟨hal-02957135v3⟩
465 Consultations
654 Téléchargements

Partager

More