Labeling the Features Not the Samples: Efficient Video Classification with Minimal Supervision
This addresses the problem of efficient video classification for applications with scarce labeled data, offering a novel approach that is incremental in its method but provides strong practical gains.
The paper tackles the problem of video classification with minimal supervision by proposing a feature selection method that requires only knowledge of whether features have stronger average values in positive versus negative samples, which can be estimated with as few as one labeled sample per class. The method achieves superior speed and performance compared to established approaches like AdaBoost, Lasso, and SVM, especially with very limited training data.
Feature selection is essential for effective visual recognition. We propose an efficient joint classifier learning and feature selection method that discovers sparse, compact representations of input features from a vast sea of candidates, with an almost unsupervised formulation. Our method requires only the following knowledge, which we call the \emph{feature sign}---whether or not a particular feature has on average stronger values over positive samples than over negatives. We show how this can be estimated using as few as a single labeled training sample per class. Then, using these feature signs, we extend an initial supervised learning problem into an (almost) unsupervised clustering formulation that can incorporate new data without requiring ground truth labels. Our method works both as a feature selection mechanism and as a fully competitive classifier. It has important properties, low computational cost and excellent accuracy, especially in difficult cases of very limited training data. We experiment on large-scale recognition in video and show superior speed and performance to established feature selection approaches such as AdaBoost, Lasso, greedy forward-backward selection, and powerful classifiers such as SVM.