Support vector machine classifiers for sequential decision problems

摘要: Classification problems in critical applications such as health care or security often require very high reliability because of the costs errors. In order to achieve this reliability, systems use sequential inspections, where additional data can be collected resolve ambiguous test cases. It is impractical costly collect on every sample, so one must find identify a policy that selects which samples need further examination. paper, we present theory for designing support vector machine classifiers include option delay decision and information. We convex programming formulation training classifiers, define fast coordinate ascent algorithm solve dual optimization problem. The performance resulting evaluated suite involving detection malignancies hyperspectral measurements colon polyps during colonoscopies.