QIM: Quantifying Hyperparameter Importance for Deep Learning

Abstract : Recently, Deep Learning (DL) has become super hot because it achieves breakthroughs in many areas such as image processing and face identification. The performance of DL models critically depend on hyperparameter settings. However, existing approaches that quantify the importance of these hyperparameters are time-consuming.In this paper, we propose a fast approach to quantify the importance of the DL hyperparameters, called QIM. It leverages Plackett-Burman design to collect as few as possible data but can still correctly quantify the hyperparameter importance. We conducted experiments on the popular deep learning framework – Caffe – with different datasets to evaluate QIM. The results show that QIM can rank the importance of the DL hyperparameters correctly with very low cost.
Type de document :
Communication dans un congrès
Guang R. Gao; Depei Qian; Xinbo Gao; Barbara Chapman; Wenguang Chen. 13th IFIP International Conference on Network and Parallel Computing (NPC), Oct 2016, Xi'an, China. Springer International Publishing, Lecture Notes in Computer Science, LNCS-9966, pp.180-188, 2016, Network and Parallel Computing. 〈10.1007/978-3-319-47099-3_15〉
Liste complète des métadonnées

Littérature citée [17 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-01648007
Contributeur : Hal Ifip <>
Soumis le : vendredi 24 novembre 2017 - 16:49:19
Dernière modification le : vendredi 24 novembre 2017 - 16:50:58

Fichier

 Accès restreint
Fichier visible le : 2019-01-01

Connectez-vous pour demander l'accès au fichier

Licence


Distributed under a Creative Commons Paternité 4.0 International License

Identifiants

Citation

Dan Jia, Rui Wang, Chengzhong Xu, Zhibin Yu. QIM: Quantifying Hyperparameter Importance for Deep Learning. Guang R. Gao; Depei Qian; Xinbo Gao; Barbara Chapman; Wenguang Chen. 13th IFIP International Conference on Network and Parallel Computing (NPC), Oct 2016, Xi'an, China. Springer International Publishing, Lecture Notes in Computer Science, LNCS-9966, pp.180-188, 2016, Network and Parallel Computing. 〈10.1007/978-3-319-47099-3_15〉. 〈hal-01648007〉

Partager

Métriques

Consultations de la notice

167