Compressive Sensing Evaluation Strategy For Prostate Cancer Classification

摘要

Compressive sensing (CS) has recently received a great deal of attention in many of real life applications. The advantage is based on its ability to utilize the sparse representation of the data. To find a sparse signal, the signal will be coded using a sensing matrix with specific properties. There are many sensing matrices that have been proposed in this field, however there is no particular matrix that is good for all applications. It is depending on the application domain, data dimensions, coding time, and of course the accuracy of recovery. In this work, different sensing matrices have been applied and compared on prostate cancer classification. The prostate cancer data has a very high number of features, so CS strategy would be a perfect for this kind of applications. The results show that the sensing matrix plays an essential role in maximizing the margin, which controls the classifier performance.