CAREER: An Ensemble of Classifiers Based Approach for Incremental Learning

职业：基于分类器集成的增量学习方法

• 批准号: 0239090
• 负责人: Robi Polikar
• 金额: $ 39.99万
• 依托单位: Rowan University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2003
• 资助国家: 美国
• 起止时间: 2003-03-01 至 2009-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0239090&HistoricalAwards=false
• 关键词: CAREER; Ensemble; Classifiers; Based; Approach

This proposal outlines a set of research and educational endeavors as part of the PI's career development plan to establish a Pattern Recognition and Machine Learning Laboratory at Rowan University. The research will address the problem of incremental learning of new in-formation without losing previously acquired knowledge, using an ensemble of classifiers approach. It will also addresses related issues that can benefit from an ensemble approach, such as estimating classification confidence and learning in non-stationary environments.The research will be tightly integrated into a variety of traditional and nontraditional educational activities. Traditional activities include developing undergraduate and graduate courses, laboratory exercises and seminars, while nontraditional activities include forming integrated research and learning communities (IRLCs) to introduce research based education to undergraduate curriculum. These communities will consist of students of all classifications, freshman through graduate (vertical integration) with a conscious effort of integrating diverse gender, ethnic, and racial backgrounds (horizontal integration). Communities will be working on different aspects of the project, with individual members responsible of tasks that are commensurate with their intellectual and academic skills. These activities will develop research, analytical thinking, problem solving, team building and communication skills of the community members, while providing maximum multidisciplinary research exposure to both undergraduate and graduate students.

该提案概述了一系列研究和教育工作，作为PI职业发展计划的一部分，在罗文大学建立模式识别和机器学习实验室。该研究将解决新信息的增量学习问题，而不会丢失以前获得的知识，使用集成的分类方法。它还将解决可以从集成方法中受益的相关问题，例如估计分类置信度和在非平稳环境中的学习，该研究将紧密结合到各种传统和非传统的教育活动中。传统的活动包括开发本科和研究生课程，实验室练习和研讨会，而非传统的活动包括形成综合研究和学习社区（IRLCs），以引入基于研究的教育本科课程。这些社区将包括所有类别的学生，大一到研究生（垂直整合），有意识地努力整合不同的性别，民族和种族背景（横向整合）。社区将致力于项目的不同方面，个人成员负责与其智力和学术技能相称的任务。这些活动将发展研究，分析思维，解决问题，团队建设和社区成员的沟通技巧，同时提供最大的多学科研究接触本科生和研究生。