Learning to hash is generating similarity-preserving binary representations of images, which is, among others, an efficient way for fast image retrieval. Two-step hashing has become a common approach because it simplifies the learning by separating binary code inference from hash function training. However, the binary code inference typically leads to an intractable optimization problem with binary constraints. Different relaxation methods, which are generally based on complicated optimization techniques, have been proposed to address this challenge. In this paper, a simple relaxation scheme based on the projected gradient is proposed. To this end in each iteration, we try to update the optimization variable as if there is no binary constraint and then project the updated solution to the feasible set. We formulate the projection step as fining closet binary matrix to the updated matrix and take advantage of the closed-form solution for the projection step to complete our learning algorithm. Inspired by opposition-based learning, pairwise opposite weights between data points are incorporated to impose a stronger penalty on data instances with higher misclassification probability in the proposed objective function. We show that this simple learning algorithm leads to binary codes that achieve competitive results on both CIFAR-10 and NUS-WIDE datasets compared to state-of-the-art benchmarks.
Feature vectors provided by pre-trained deep artificial neural networks have become a dominant source for image representation in recent literature. Their contribution to the performance of image analysis can be improved through finetuning. As an ultimate solution, one might even train a deep network from scratch with the domain-relevant images, a highly desirable option which is generally impeded in pathology by lack of labeled images and the computational expense. In this study, we propose a new network, namely KimiaNet, that employs the topology of the DenseNet with four dense blocks, fine-tuned and trained with histopathology images in different configurations. We used more than 240,000 image patches with 1000×1000 pixels acquired at 20× magnification through our proposed "highcellularity mosaic" approach to enable the usage of weak labels of 7,126 whole slide images of formalin-fixed paraffin-embedded human pathology samples publicly available through the The Cancer Genome Atlas (TCGA) repository. We tested KimiaNet using three public datasets, namely TCGA, endometrial cancer images, and colorectal cancer images by evaluating the performance of search and classification when corresponding features of different networks are used for image representation. As well, we designed and trained multiple convolutional batch-normalized ReLU (CBR) networks. The results show that KimiaNet provides superior results compared to the original DenseNet and smaller CBR networks when used as feature extractor to represent histopathology images.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.