Using ROC and Unlabeled Data for Increasing Low Shot Transfer Learning Classification Accuracy

Convolutional Neural Networks can recognize only the classes that they are trained for . When using them for classification, any candidate image will be placed in one of the available classes . We propose a low-shot classifier which can serve as the top layer to any existing CNN that the feature extractor was already trained . Using a limited amount of labeled data for the type of images which need to be specifically classified along with unlabeled data for all other images, a unique target matrix and a Receiver Operator Curve (ROC) criterion, we are able to increase identification accuracy by up to 30% for the images that do not belong to any specific classes, while retaining the ability to identify images that belong to the specific classes of interest . The low-shooter classifier can be added to any CNN that was previously trained. We propose to use a low shot classifier to help us to identify an image that was already identified by an algorithm that was trained for the image that we have already been trained. It can be used

Links: PDF - Abstract

Code :

https://github.com/skasapis/ROCUnlabeledClassification

Keywords : trained - classes - images - shot - data -

Leave a Reply

Your email address will not be published. Required fields are marked *