This paper aims to increase the accuracy of Kazakh handwriting text recognition (KHTR) using the generative adversarial network (GAN), where a handwriting word image generator and an image quality discriminator are constructed. In order to obtain a high-quality image of handwritten text, the multiple losses are intended to encourage the generator to learn the structural properties of the texts. In this case, the quality discriminator is trained on the basis of the relativistic loss function. Based on the proposed structure, the resulting document images not only preserve texture details but also generate different writer styles, which provides better OCR performance in public databases. With a self-created dataset, images of different types of handwriting styles were obtained, which will be used when training the network. The proposed approach allows for a character error rate (CER) of 11.15% and a word error rate (WER) of 25.65%.
The ability of the end user to work with a large amount of data from a large number of heterogeneous sources and at the same time get an effective result from the work is carried out through the use of graphical web interfaces built on the basis of XML technologies that allow displaying any structure of a file presented in XML format. As a data exchange method between applications on the Web, XML still lacks capabilities for identification of web resources and a system that uses them, and capabilities to express the knowledge provided by XLM documents. In this study, a web interface has been developed (a web-based server application), as an XML records editor that provides display forms for the creation and editing of XML documents and is able to adapt to the internal resources of the system used. The technology is based on the XSD data set schema transformation by the way of XSLT transformations. Screen forms are generated on the server side and are provided to the user with all the necessary tools for correct input and/or editing of heterogeneous data. A distinguishing characteristic of this technology is the ability to display both properly and improperly formed XML data. The developed graphical interface allows any application to automatically exchange and read information from other applications without human intervention, which significantly improves performance and ease of use. This software solution could be used both as an independent data building and editing module presented in the XML format, and as a built-in module plugged into various server software for heterogeneous information management systems
The paper is devoted to solve the pattern recognition problem with incomplete learning data. The solution method, which combines similarity graph with Laplacian Regularization and collective clustering is proposed. The low-rank decomposition of co-association matrix for cluster ensemble is used, which allows to speed up the computations and keep memory. Experimental results on test tasks and on real hyperspectral image demonstrate the effectiveness of proposed method, including with noisy data.
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.