Multiple instance learning via margin maximization

详细信息    查看全文

• 作者：O. Erhun Kundakcioglu ; Onur Seref ; Panos M. Pardalos
• 关键词：Multiple instance learning ; Support vector machines ; Branch and bound ; Pattern classification ; Object recognition ; Drug activity prediction
• 刊名：Applied Numerical Mathematics
• 出版年：2010
• 出版时间：April 2010
• 年：2010
• 卷：60
• 期：4
• 页码：358-369
• 全文大小：215 K

文摘

In this paper, we consider the classification problem within the multiple instance learning (MIL) context. Training data is composed of labeled bags of instances. Despite the large number of margin maximization based classification methods, there are only a few methods that consider the margin for MIL problems in the literature. We first formulate a combinatorial margin maximization problem for multiple instance classification and prove that it is -hard. We present a way to apply the kernel trick in this formulation for classifying nonlinear multiple instance data. We also propose a branch and bound algorithm and present computational results on publicly available benchmark data sets. Our approach outperforms a leading commercial solver in terms of the best integer solution and optimality gap in the majority of image annotation and molecular activity prediction test cases.