Abstract
Due to the limitation of data annotation and the ability of dealing with label-efficient problems, active learning has received lots of research interest in recent years. Most of the existing approaches focus on designing a different selection strategy to achieve better performance for special tasks; however, the performance of the strategy still needs to be improved. In this work, we focus on improving the performance of active learning and propose a loss-based strategy that learns to predict target losses of unlabeled inputs to select the most uncertain samples, which is designed to learn a better selection strategy based on a double-branch deep network. Experimental results on two visual recognition tasks show that our approach achieves the state-of-the-art performance compared with previous methods. Moreover, our approach is also robust to different network architectures, biased initial labels, noisy oracles, or sampling budget sizes, and the complexity is also competitive, which demonstrates the effectiveness and efficiency of our proposed approach.
Highlights
In recent years, due to the strong ability of feature extraction, big data, and advanced hardware, deep learning (DL) has received lots of research interests and plays an important role in many fields, especially in computer vision, such as image classification,[1,2] object detection,[3,4] and image segmentation.[5]
We focus on the performance improvement of AL and to eliminate the limitations of existing methods, we propose a novel approach for AL, which is called double-branch deep network active learning (DNAL)
We find that the performances of variational adversarial approach for active learning (VAAL), Core-set, and Entropy methods are very close among the stages, this might be because there are only 10 categories on the Canadian Institute for Advanced Research (CIFAR)-10 data set, and 5% budget selection at each stage might be enough for such methods
Summary
Due to the strong ability of feature extraction, big data, and advanced hardware, deep learning (DL) has received lots of research interests and plays an important role in many fields, especially in computer vision, such as image classification,[1,2] object detection,[3,4] and image segmentation.[5]. We proposed a novel framework of learning loss by a double-branch deep network, which is designed to learn the uncertainty through learning a loss L during training, and we select most K informative samples to be labeled based on the uncertainty of the value of the loss during sampling stage, which enables the system to learn a better sampling strategy. Double networks have been adopted in some other tasks,[22,23] it is first introduced for AL here, inspired by Deep Q-Network,[22] instead of carefully designing a strategy based on comparing a pair of samples to learn the loss-prediction loss,[20] we proposed a novel framework of learning loss by combining two deep networks. Training process based on Double-branch deep Network for Active Learning(DNAL)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
