项目名称: 压缩感知和稀疏优化中的非凸优化算法设计
项目编号: No.11471205
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 数理科学和化学
项目作者: 葛冬冬
作者单位: 上海财经大学
项目金额: 60万元
中文摘要: 大数据时代对海量数据的压缩,储存和恢复提出了更高的要求。因此,根据大数据的稀疏性特点,设计相关的模型和算法,成为当前在工程的压缩感知,经济与管理学的资产管理与优化,统计学的桥估计量的重要问题。也成为运筹与优化理论中一个新的方向,稀疏优化。本申请书提出了对此类问题常见的一个模型,L2-Lp模型的深入理论探讨。并为此类非凸优化问题如何寻求近似最优解(KKT近似点),提出了一些新的不同于以往的算法,并对算法的具体实施中的技巧进行了探讨,也对此类稀疏算法在实际问题中的可能应用做了前景预测。
中文关键词: 最优化理论;稀疏优化;压缩感知;非线性规划;内点算法
英文摘要: In Big Data Era, the demand for compressing, saving and recovering large amount of data has been becoming stronger and stronger. Designing efficient models and algorithms based on the sparsity of big data has played a key role in many problems, such as compressed sensing in engineering, protofolio management and optimization in economics and business management, bridge estimator in statistics. It has been developed to an important new field in operations research and optimization theory: sparse optimization. Our proposal plans a further discussion on a classical model in this subject: L2-Lp Model. In this proposal we provide some potentially different approaches, try to deliver detailed possible techniques in implementation, and forcast its possible applications in real situation.
英文关键词: Optimization Theory;Sparse Optimization;Compressed sensing;nonlinear programming;interior point algorithm