项目名称: 面向大数据的哈希学习理论与应用
项目编号: No.61472182
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 其他
项目作者: 李武军
作者单位: 南京大学
项目金额: 80万元
中文摘要: 大数据学习已经成为大数据研究的核心问题之一,而哈希学习通过将数据表示成二进制码的形式,能大大减少数据的存储和通信开销,从而大大提高大数据学习系统的效率。因此,哈希学习于近几年迅速发展成为机器学习领域和大数据学习领域的一个研究热点,并被广泛应用于数据挖掘、模式识别、信息检索等领域。本项目通过对已有哈希学习方法的详细调研,在申请人课题组近几年的预研和初步探索基础上,从哈希学习的本质问题入手,重点研究哈希学习的模型构建、参数优化、量化方法以及性能评估等方面的内容,突破理论和计算等关键技术,构建一套既具有理论创新又具有实际应用价值的哈希学习工具包,发表高水平学术论文,并产生具有自主知识产权的专利技术,为我国在哈希学习和大数据学习领域的发展提供技术储备,并带动相关应用领域的发展。
中文关键词: 哈希学习;大数据;人工智能;机器学习;数据挖掘
英文摘要: Big data machine learning (BDML) is one of the core problems in big data research. By representing the data with binary codes, Learning to Hash (LH) can dramatically reduce the cost for storage and communication, which can greatly improve the efficiency of the BDML systems. Hence, learning to hash has become a very hot research topic in machine learning and BDML,with wide application in data mining, pattern recognition, and information retrieval and so on. Based on the pre-study conducted in the lab of the applier, this project tries to focus on the essential problems underlying LH.The key aspects pursued by this project include model building, parameter optimization and learning, quatization strategy, and performance evaluation. The goal of this project is to build a toolbox with both theoretical and practical contributions for LH. The expected outcomes include high-quality papers on top international journals and conferences, and patent application.
英文关键词: Learning to Hash;Big Data;Artificial Intelligence;Machine Learning;Data Mining