Recently, relevance feedback (RF) in content-based image retrieval (CBIR) has been implemented as an online binary classifier to separate the positive samples from the negative samples, where both sets of samples are labeled by the user. In many applicat
