探索由电池供电的移动设备上深度卷积神经网络的训练：设计与应用（CS LG）---刘持诚

移动设备上快速发展的智能应用利用预训练的深度学习模型进行推理。然而，这些模型通常不会在此后进行更新。这给适应新的数据分布留下了很大的差距。在本文中，我们进一步将深度神经网络的训练纳入到电池供电的移动设备上。我们从性能和隐私方面发现了阻碍动态移动环境下有效学习的几个挑战。我们将问题重构为度量学习来解决超拟合问题，并在内存约束下通过数据解析扩大样本空间。我们还使该方案能够抵御侧通道攻击和运行时的波动。我们进行了一个基于深度行为认证的案例研究。实验表明，在三个公共数据集上的准确率超过 95%，在较少的数据下多类分类的准确率提高了 15%，并且对暴力攻击和侧通道攻击的鲁棒性分别达到 99% 和 90% 的成功率。我们展示了使用移动 CPU 进行训练的可行性，训练 100 个 epoch 只需不到 10 分钟，并且通过特征转移可以提升 3 - 5 倍。最后，我们对内存、能量和计算开销进行了剖析。我们的研究结果表明，训练比看视频消耗的能量要低，比玩游戏消耗的能量略高。

原文题目：Explore Training of Deep Convolutional Neural Networks on Battery-powered Mobile Devices: Design and Application

原文：The fast-growing smart applications on mobile devices leverage pre-trained deep learning models for inference. However, the models are usually not updated thereafter. This leaves a big gap to adapt the new data distributions. In this paper, we take a step further to incorporate training deep neural networks on battery-powered mobile devices. We identify several challenges from performance and privacy that hinder effective learning in a dynamic mobile environment. We re-formulate the problem as metric learning to tackle overfitting and enlarge sample space via data paring under the memory constraints. We also make the scheme robust against side-channel attacks and run-time fluctuations. A case study based on deep behavioral authentication is conducted. The experiments demonstrate accuracy over 95% on three public datasets, a sheer 15% gain from multi-class classification with less data and robustness against brute-force and side-channel attacks with 99% and 90% success, respectively. We show the feasibility of training with mobile CPUs, where training 100 epochs takes less than 10 mins and can be boosted 3-5 times with feature transfer. Finally, we profile memory, energy and computational overhead. Our results indicate that training consumes lower energy than watching videos and slightly higher energy than playing games.

原文作者：Cong Wang, Yanru Xiao, Xing Gao, Li Li, Jun Wang

原文地址：https://arxiv.org/abs/2005.12901