Management of Mobile Phone Sensing via Sparse Learning

通过稀疏学习管理手机传感

• 批准号: 1818904
• 负责人: Yingbin Liang
• 金额: $ 31.44万
• 依托单位: Ohio State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-10-01 至 2020-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1818904&HistoricalAwards=false
• 关键词: Management; Mobile; Phone; Sensing; via

Today's smartphones carry various built-in sensors such as GPS, accelerometers, and cameras to acquire information. These sensors also enable smartphones to collectively accomplish various sensing services on demand, for example environmental monitoring and traffic measurement and control over certain geographical areas. This type of mobile phone sensing allows ubiquitous coverage due to mobility of smartphones. It is also cost effective, as it does not need a dedicated sensing infrastructure and can adapt more flexibly to different sensing demands. The major design issue for mobile phone sensing systems is to minimize battery usage/energy consumption and reduce interference with primary voice and data services of smartphones while maintaining the desired quality of sensing services. This requires activating only a subset of smartphones that best serve the sensing task. Such sensor selection problems are very challenging for large-scale networks, as they are combinatorial in nature and intractable in general. The goal of this project is to design a comprehensive sparse learning based methodology, which converts challenging combinatorial sensor selection problems to sparsity-encouraged convex optimization problems that can be solved efficiently. The research agenda will include development of sparse learning approaches in view of sensor collaboration structures, design of incentive-based mechanisms to fulfill sensing demands as well as individual sensors profits, and design of dynamic management policies that adapt to time-varying environments and incentives of sensors. Success of this project will greatly advance sensor management designs and potentially significantly improve efficiency of practical mobile phone sensing systems. This project will offer hands-on training opportunities to students from diverse backgrounds.

今天的智能手机携带各种内置传感器，如GPS，加速度计和摄像头来获取信息。这些传感器还使智能手机能够按需共同完成各种传感服务，例如环境监测和某些地理区域的交通测量和控制。由于智能手机的移动性，这种类型的移动的电话感测允许无处不在的覆盖。它还具有成本效益，因为它不需要专用的传感基础设施，并且可以更灵活地适应不同的传感需求。移动的电话感测系统的主要设计问题是最小化电池使用/能量消耗，并减少对智能电话的主要语音和数据服务的干扰，同时保持感测服务的期望质量。这需要仅激活最适合传感任务的智能手机的子集。这样的传感器选择问题对于大规模网络来说是非常具有挑战性的，因为它们本质上是组合的，并且通常是棘手的。这个项目的目标是设计一个全面的稀疏学习为基础的方法，将具有挑战性的组合传感器选择问题，稀疏鼓励凸优化问题，可以有效地解决。研究议程将包括稀疏学习方法的发展，鉴于传感器的协作结构，设计基于激励的机制，以满足传感需求以及个人传感器的利润，并设计动态管理策略，以适应时变的环境和传感器的激励。该项目的成功将极大地推进传感器管理设计，并有可能显著提高实际移动的电话传感系统的效率。该项目将为来自不同背景的学生提供实践培训机会。

项目成果

Analysis of Q-learning with Adaptation and Momentum Restart for Gradient Descent

• DOI: 10.24963/ijcai.2020/422
• 发表时间: 2020-07
• 作者: Chuhan Wu;Fangzhao Wu;Tao Qi;Yongfeng Huang

Convergence of cubic regularization for nonconvex optimization under KL property

KL性质下非凸优化的三次正则化收敛

• 发表时间: 2018
• 期刊: Proc. Advances in Neural Information Processing Systems (NeurIPS
• 作者: Zhou, Y;Wang, Z;Liang, Y.
• 通讯作者: Liang, Y.

Median-Truncated Nonconvex Approach for Phase Retrieval With Outliers

• DOI: 10.1109/tit.2018.2847695
• 发表时间: 2016-03
• 期刊: IEEE Transactions on Information Theory
• 影响因子: 2.5
• 作者: Huishuai Zhang;Yuejie Chi;Yingbin Liang

History-Gradient Aided Batch Size Adaptation for Variance Reduced Algorithms

• 发表时间: 2019-10
• 作者: Kaiyi Ji;Zhe Wang-;Bowen Weng;Yi Zhou;W. Zhang;Yingbin Liang

Critical Points of Linear Neural Networks: Analytical Forms and Landscape Properties

• 发表时间: 2017-10
• 作者: Yi Zhou;Yingbin Liang