[2005.10524] HyperSTAR: Task-Aware Hyperparameters for Deep Networksopen searchopen navigation menucontact arXivarXiv Twitter

While deep neural networks excel in solving visual recognition tasks, they require significant effort to find hyperparameters that make them work optimally. Hyperparameter Optimization (HPO) approaches have automated the process of finding good hyperparameters but they do not adapt to a given task (task-agnostic), making them computationally inefficient. To reduce HPO time, we present HyperSTAR (System for Task Aware Hyperparameter Recommendation), a task-aware method to warm-start HPO for deep neural networks. HyperSTAR ranks and recommends hyperparameters by predicting their performance conditioned on a joint dataset-hyperparameter space. It learns a dataset (task) representation along with the performance predictor directly from raw images in an end-to-end fashion. The recommendations, when integrated with an existing HPO method, make it task-aware and significantly reduce the time to achieve optimal performance. We conduct extensive experiments on 10 publicly available large-scale

4 mentions: @VMFragoso@kuritateppei
Date: 2020/05/23 02:21

Referring Tweets

@kuritateppei 未知のデータセットに対するハイパーパラメータ(学習率、Optimizer、Augmentation…)の推薦、ランク付けを行う。特定のデータセットとハイパラ構成を入力したときに、そのパフォーマンス予測器をEnd2Endで学習したのがキモ。いわゆるメタ学習。 CVPR2020オーラル。 t.co/8acR7h6pNQ t.co/a6njADKW3K
@VMFragoso Overwhelmed by finding optimal hyper-parameters of a CNN for a task? Check out our HyperSTAR CVPR paper, a new method that: ☑️ “Understands” a task ☑️ Recommends hyper-parameters for it ☑️ Accelerates a hyper-parameter optimization algorithm 👉 t.co/xpbU0DqLE5 #CVPR2020 t.co/r7ViH1XqIJ

Related Entries

Read more [2002.10631] Batch norm with entropic regularization turns deterministic autoencoders into generativ...
0 users, 3 mentions 2020/02/27 00:52
Read more [2002.12213] Meta-Transfer Learning for Zero-Shot Super-Resolutioncontact arXivarXiv Twitter
0 users, 4 mentions 2020/03/01 17:20
Read more [2003.08237] Fixing the train-test resolution discrepancy: FixEfficientNetcontact arXivarXiv Twitter
0 users, 3 mentions 2020/03/23 06:52
Read more [2004.08566] On the Synergies between Machine Learning and Stereo: a Surveyopen searchopen navigatio...
0 users, 2 mentions 2020/04/24 03:51
Read more [2004.09754] The 1st Agriculture-Vision Challenge: Methods and Resultsopen searchopen navigation men...
0 users, 2 mentions 2020/04/27 05:21