• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>IEEE Transactions on Parallel and Distributed Systems >SmartTuning: Selecting Hyper-Parameters of a ConvNet System for Fast Training and Small Working Memory
【24h】

SmartTuning: Selecting Hyper-Parameters of a ConvNet System for Fast Training and Small Working Memory

机译：SmartTuning：选择ConvNet系统的超参数，以进行快速培训和小型工作记忆

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

It is desirable to deploy a ConvNet system with high inference accuracy, as well as fast training and small inference memory. However, existing approaches to hyper-parameter tuning only focus on high accuracy. Although achieving high accuracy, tuning poorly can significantly increase the performance burden, and thus degrade the overall performance of a ConvNet system. In this article, we propose SmartTuning, an approach to identifying the hyper-parameters of a ConvNet system for high training speed and small working memory, with the restriction of high inference accuracy. The key idea of SmartTuning is to build a new performance model for a ConvNet system, and to integrate Bayesian Optimization to learn the relationship between the overall performance and the hyper-parameters of a ConvNet system. In this way, SmartTuning can balance inference accuracy, training speed and inference memory usage during the tuning process, and thus maximizes the overall performance of a ConvNet system. Our experiments show that SmartTuning can stably identify the hyper-parameter sets that offer very close accuracy with faster training speed (i.e., 7x-11x over MNIST and 2x-3x over CIFAR-10) and much less inference memory usage (i.e., 17x-23x over MNIST and 4x-9x over CIFAR-10), compared with existing tuning approaches.
机译：期望以高推理精度和快速训练和小推理存储器部署ConvNet系统。但是，超参数调谐的现有方法仅关注高精度。虽然实现了高精度，但调整不良会显着提高性能负担，从而降低了Convnet系统的整体性能。在本文中，我们提出了SmartTuning，一种方法来识别Convnet系统的高参数，用于高训练速度和小型工作记忆，具有高推理精度的限制。 SmartTuning的关键思想是为Convnet系统构建一个新的性能模型，并集成贝叶斯优化，以了解Convnet系统的整体性能和超参数之间的关系。通过这种方式，SmartTuning可以在调整过程中平衡推理准确性，训练速度和推理内存使用量，从而最大限度地提高了ConvNet系统的整体性能。我们的实验表明，SmartTuning可以稳定地识别超参数集，以更快的训练速度（即7x-11x OVER MNIST，在CIFAR-10上为2x-3x），并且推断内存使用量更少（即，17x-与现有的调谐方法相比，在CNIST和4x-9x上的MNIST和4x-9x上。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号