Approximating Optimal Solutions Using Polynomial-Time Learning Machines and Applications in Intelligent Network Management

使用多项式时间学习机逼近最佳解决方案及其在智能网络管理中的应用

• 批准号: 9502518
• 负责人: Chuanyi Ji
• 金额: $ 13.5万
• 依托单位: Rensselaer Polytechnic Institute
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1995
• 资助国家: 美国
• 起止时间: 1995-08-01 至 1998-10-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9502518&HistoricalAwards=false
• 关键词: Approximating; Optimal; Solutions; Using; Polynomial

The approach of this Career Research Project is to apply weak classifiers in machine learning systems to produce better learning rules. Weak classifiers are those that do not allow correct concept identification all the time, but that work better than random guesses. Alone, they are not good enough for applications, but in combinations, they may be useful. There are two major considerations - efficiency of learning and generalization error. Within this approach, the goal of this research is to rigorously investigate the fundamental problems of learning systems in supervised learning performance and efficiency (2) to apply adaptive learning machines to intelligent network management.

这个职业研究项目的方法是应用 机器学习系统中的弱分类器， 更好的学习规则。 弱分类器是那些 不允许始终正确的概念识别，但 比随机猜测更有效 单独使用它们还不足以满足应用要求，但将它们结合起来可能会很有用。 有两个主要的考虑因素-效率 学习和泛化的错误。 在这种方法中，本研究的目标是严格调查学习的基本问题 监督学习系统的性能和效率 （二） 应用自适应学习 机器 智能网络管理。