摘要
为提高基于选择性集成的表情识别方法的泛化能力,降低其预测成本,提出一种基于带极值扰动的简化粒子群优化和选择性集成的表情识别方法。采用简化粒子群优化算法对基分类器进行选择集成,将基于选择性集成的表情识别问题转化为半定规划问题进行解决。对比实验结果表明,该方法比传统方法有更好的识别率和鲁棒性。
To improve the generalization ability and reduce the prediction cost of the facial recognition method based on selective ensemble,an emotion recognition method based on extremum disturbed and simplified particle swarm optimization（tsPSO）algorithm and selective ensemble was proposed.Simplified particle swarm optimization algorithm was used to select the base classifiers and ensemble,the issue of facial expression recognition based on selective ensemble was transformed to a semi-definite programming problem.The comparative experimental results show that the proposed method is superior to the traditional methods and it is more robust.
出处
《计算机工程与设计》
北大核心
2017年第3期773-778,共6页
Computer Engineering and Design
基金
韩国科学与信息科技未来规划部2013年ICT研发基金项目(10039149)
重庆市自然科学基金项目(CSTC
2007BB2445)
2015年重庆市研究生科研创新基金项目(CYS15174)
关键词
集成学习
选择性集成
半定规划
简化粒子群优化
表情识别
ensemble learning
selective ensemble
semi-definite programming
simplified particle swarm optimization
facial expression recognition
