Search

To identify choroidal characteristics associated with susceptibility to development of naturally occurring and experimentally induced myopia.We compared choroidal properties between pigmented and albino guinea pig (GP) strains. Biometry, cycloplegic refractive error (RE), and eye wall sublayer thickness were measured from 171 GPs at postnatal day (P)6, 14, and 28. Forty-three P14 GPs underwent two-week monocular form-deprivation myopia (FDM). En face images of choroidal vasculature were obtained with a customized swept-source optical coherence tomography. Multivariate regression analyses were performed, with P28 RE as the outcome and P14 choroidal thickness (ChT) as the main predictor variable. Proteomic analysis was performed on choroidal tissue from P14 albino and pigmented GPs.At P14, RE was correlated with thickness of the choroid (β = 0.06), sclera (β = 0.12), and retina (β = 0.27; all P0.001). P14 ChT was correlated with P28 RE both with (β = 0.06, P = 0.0007) and without FDM (β = 0.05, P = 0.008). Multivariate regression analysis, taking into account FDM (versus physiological growth) and strain, revealed that for every 10-µm greater ChT at P14, P28 RE was 0.50D more positive (P = 0.005, n = 70). En face images of choroidal sublayers showed that albino choroids were relatively underdeveloped, with frequent avascular regions. Consistent with this finding, proteomic analysis suggested abnormalities of the nitric oxide system in the albino GP choroid.Current results are consistent with the notion that greater ChT could protect from or delay the onset of myopia, while lower ChT is associated with greater susceptibility to myopia development. The underlying mechanism could be related to dysfunction of the choroidal vascular system.

The rapid development of industry enterprises, the large amount of data generated by these originalities and the exponential growth of industrial business website are the causes that lead to different types of big data and data stream problem. There are many stream data mining algorithms for classification and clustering with their specific properties and significance key features. Ensemble classifiers help to improve the best predictive performance results among these up-to-date algorithms. In ensemble methods, different kinds of classifiers and clusters are trained rather than training single classifier. Their prediction machine learning results are combined to a voting schedule. This paper presented a framework for stream data mining by taking the benefits of assembling technology based on miss classification stream data. Experiments are carried out with real world data streams. The experimental performance results are compared with the modern popular ensemble techniques such as Boosting and Bagging. The increasing in accuracy rate and the reducing in classification time can be seen from the test results.

Automatic machine-printed Optical Characters or texts Recognizers (OCR) are highly desirable for a multitude of modern IT applications, including Digital Library software. However, the state of the art OCR systems cannot do for Myanmar scripts as the language poses many challenges for document understanding. Therefore, the authors design an Optical Character Recognition System for Myanmar Printed Document (OCRMPD), with several proposed techniques that can automatically recognize Myanmar printed text from document images. In order to get more accurate system, the authors propose the method for isolation of the character image by using not only the projection methods but also structural analysis for wrongly segmented characters. To reveal the effectiveness of the segmentation technique, the authors follow a new hybrid feature extraction method and choose the SVM classifier for recognition of the character image. The proposed algorithms have been tested on a variety of Myanmar printed documents and the results of the experiments indicate that the methods can increase the segmentation accuracy as well as recognition rates.

Keyword-based search is popularized in Information Retrieval (IR) communities and Internet search engines on the Web. Nowadays, social networking, micro-blogging, and other data-driven websites store large amounts of information in relational databases. But searching in databases users need to know a database schema and a query language. The usability of Database is severely limited unless easier ways to access databases are developed. While traditional database management systems offer powerful query languages, they do not allow keyword-based search. Keyword search techniques on the Web cannot directly be applied to databases because the data on the Internet and database are in different forms. Therefore, we propose a search system for relational data with our proposed methods that can support keyword-based search over relational database. The main contributions of this system compose of the direct indexing of records in database, the schema based queries for relevant answers, and ranking of the relevant answer sets. As long as the database table records can be extended, this system can be easily extendable for further searching records from tables.