Event detection of different English data sources based on transfer learning

2021 ◽  
pp. 1-11
Author(s):  
Yanan Huang ◽  
Yuji Miao ◽  
Zhenjing Da
Keyword(s):  
Transfer Learning ◽  
Event Detection ◽  
Visual Analysis ◽  
Learning Algorithm ◽  
Data Sources ◽  
Data Set ◽  
Data Source ◽  
Single Data Source ◽  
The Difference ◽  
Single Data

The methods of multi-modal English event detection under a single data source and isomorphic event detection of different English data sources based on transfer learning still need to be improved. In order to improve the efficiency of English and data source time detection, based on the transfer learning algorithm, this paper proposes multi-modal event detection under a single data source and isomorphic event detection based on transfer learning for different data sources. Moreover, by stacking multiple classification models, this paper makes each feature merge with each other, and conducts confrontation training through the difference between the two classifiers to further make the distribution of different source data similar. In addition, in order to verify the algorithm proposed in this paper, a multi-source English event detection data set is collected through a data collection method. Finally, this paper uses the data set to verify the method proposed in this paper and compare it with the current most mainstream transfer learning methods. Through experimental analysis, convergence analysis, visual analysis and parameter evaluation, the effectiveness of the algorithm proposed in this paper is demonstrated.

scholarly journals Comparison of bibliographic data sources: Implications for the robustness of university rankings

2020 ◽  
pp. 1-34 ◽  
Author(s):  
Chun-Kai (Karl) Huang ◽  
Cameron Neylon ◽  
Chloe Brookes-Kenworthy ◽  
Richard Hosking ◽  
Lucy Montgomery ◽  
...  
Keyword(s):  
Citation Count ◽  
Data Sources ◽  
University Rankings ◽  
Multiple Sources ◽  
Data Set ◽  
Bibliographic Data ◽  
Data Source ◽  
English Speaking ◽  
International University ◽  
Single Data

Universities are increasingly evaluated on the basis of their outputs. These are often converted to simple and contested rankings with substantial implications for recruitment, income, and perceived prestige. Such evaluation usually relies on a single data source to define the set of outputs for a university. However, few studies have explored differences across data sources and their implications for metrics and rankings at the institutional scale. We address this gap by performing detailed bibliographic comparisons between Web of Science (WoS), Scopus, and Microsoft Academic (MSA) at the institutional level and supplement this with a manual analysis of 15 universities. We further construct two simple rankings based on citation count and open access status. Our results show that there are significant differences across databases. These differences contribute to drastic changes in rank positions of universities, which are most prevalent for non-English-speaking universities and those outside the top positions in international university rankings. Overall, MSA has greater coverage than Scopus and WoS, but with less complete affiliation metadata. We suggest that robust evaluation measures need to consider the effect of choice of data sources and recommend an approach where data from multiple sources is integrated to provide a more robust data set.

scholarly journals Imputation Techniques in Microsimulation

2010 ◽  
Vol 25 (3) ◽  
pp. 69-83
Author(s):  
Kum Hyun Sub
Keyword(s):  
Propensity Score ◽  
Synthetic Data ◽  
Data Sources ◽  
New Approach ◽  
Key Variables ◽  
Data Source ◽  
Policy Measures ◽  
Single Data Source ◽  
Single Data ◽  
The Individual

Microsimulation has gained attention for its use in analyzing and forecasting the individual impacts of alternative economic and social policy measures. In practice, however, microsimulation cannot be carried out from a single data source, since it requires far more information than any single data source can provide. This paper discusses ways to combine separate data sources when there are no identical key variables, using imputation techniques, to make a large but synthetic data source for microsimulation. A new approach based on propensity score matching is suggested and discussed.

scholarly journals MASHUP SCHEME DESIGN OF MAP TILES USING LIGHTWEIGHT OPEN SOURCE WEBGIS PLATFORM

2018 ◽  
Vol XLII-3 ◽  
pp. 565-571
Author(s):  
T. Hu ◽  
J. Fan ◽  
H. He ◽  
L. Qin ◽  
G. Li
Keyword(s):  
Coordinate System ◽  
Low Cost ◽  
Image Data ◽  
Source Image ◽  
Scheme Design ◽  
Paper Cover ◽  
Data Source ◽  
Single Data Source ◽  
Single Data ◽  
Loading Scheme

To address the difficulty involved when using existing commercial Geographic Information System platforms to integrate multi-source image data fusion, this research proposes the loading of multi-source local tile data based on CesiumJS and examines the tile data organization mechanisms and spatial reference differences of the CesiumJS platform, as well as various tile data sources, such as Google maps, Map World, and Bing maps. Two types of tile data loading schemes have been designed for the mashup of tiles, the single data source loading scheme and the multi-data source loading scheme. The multi-sources of digital map tiles used in this paper cover two different but mainstream spatial references, the WGS84 coordinate system and the Web Mercator coordinate system. According to the experimental results, the single data source loading scheme and the multi-data source loading scheme with the same spatial coordinate system showed favorable visualization effects; however, the multi-data source loading scheme was prone to lead to tile image deformation when loading multi-source tile data with different spatial references. The resulting method provides a low cost and highly flexible solution for small and medium-scale GIS programs and has a certain potential for practical application values. The problem of deformation during the transition of different spatial references is an important topic for further research.

scholarly journals Bees Assorter

2020 ◽  
Vol 6 (12) ◽  
pp. 61-65
Author(s):  
Himanshu Verma
Keyword(s):  
Machine Learning ◽  
Learning Algorithm ◽  
Machine Learning Algorithms ◽  
Bumble Bee ◽  
Wing Size ◽  
Qualitative And Quantitative Analysis ◽  
Data Set ◽  
Qualitative And Quantitative ◽  
The Difference

Many attempts were made to classify the bees that is bumble bee or honey bee , there have been such a large amount of researches which were made to seek out the difference between them on the premise of various features like wing size , size of bee , color, life cycle and many more. But altogether the analysis there have been either that specialize in qualitative or quantitative , but to beat this issue , thus researchers came up with an answer which might be both qualitative and quantitative analysis made to classify them. And making use of machine learning algorithm to classify them gives a lift . Now the classification would take less time as these algorithms are pretty fast and accurate . By using machine learning work is made easy . Lots of photographs had to be collected and stored for data set. And by using these machine learning algorithms we would be getting information about the bees which might be employed by researchers in further classification of bees. Manipulation of images had to be done so as on prepare them in such a way that they will be applied to the algorithms and have feature extraction done. As there have been a lot of photographs(data set) which take a lot of space and also the area in which bees were present in these photographs were too small so to accommodate it dimension reduction was done , it might not consider other images like trees , leaves , flowers which were there present in the photograph which we elect as a data set.

scholarly journals Efficient Capacity-Constrained Multicast in RC-Based P2P Networks

10.29007/dhwl ◽  
2019 ◽  
Author(s):  
Koushik Maddali ◽  
Banafsheh Rekabdar ◽  
Swathi Kaluvakuri ◽  
Bidyut Gupta
Keyword(s):  
Residue Class ◽  
P2p Networks ◽  
Multicast Protocol ◽  
Multicast Protocols ◽  
Application Level Multicast ◽  
Data Source ◽  
Single Data Source ◽  
Single Data ◽  
Capacity Constrained ◽  
Better Than

Application level multicast is independent of router infrastructure unlike router-based IP multicast. The existing DHT-based application level multicast protocols work efficiently as long as there is almost no churn; otherwise, their performances start degrading drastically, because DHT – based architecture cannot handle churn effectively. Besides, most of DHT-based multicast protocols consider single data source and do not consider peer heterogeneity. In this work, we have considered an existing non-DHT based P2P architecture, viz., Residue Class based (RC-based) architecture which has already been shown to perform much better than some well-known DHT-based architectures from the viewpoints of speed of unicast communication and churn handling. We have presented a highly efficient capacity-constrained and any source multicast protocol suitable for the RC-based P2P architecture as mentioned above.

scholarly journals Detecting abnormal fundus images by employing deep transfer learning

2019 ◽  
Author(s):  
yan yu ◽  
Changfan Wu ◽  
Xiao Chen ◽  
Xiangbing Zhu ◽  
Yinfen Hou ◽  
...  
Keyword(s):  
Transfer Learning ◽  
Learning Algorithm ◽  
Care Center ◽  
High Sensitivity ◽  
Vascular Lesion ◽  
Fundus Photography ◽  
Fundus Images ◽  
Validation Data ◽  
Data Set ◽  
Health Care Center

Abstract BackgroundTo develop and validate a deep transfer learning (DTL) algorithm in detecting abnormalities of fundus images from non-mydriatic fundus photography examination.Methods1,295 fundus images from January 2017 to December 2018 at Yijishan Hospital of Wannan Medical College were collected for developing and validating the deep transfer learning algorithm in detecting abnormal fundus images. The DTL model was developed by using 929(normal 254, abnormal 402) fundus images, including normal fundus images and abnormal fundus images, the latter including, maculopathy, optic neuropathy, vascular lesion, choroidal lesions, vitreous disease, cataract and the others. We tested our model using a subset of the publically available MESSIDOR dataset (using 366 images) and evaluate the testing performance of the DTL model for detecting abnormal fundus images. ResultsIn the internal validation data set (n=273 images), the AUC, sensitivity, accuracy and specificity of the DTL for correctly classified funds images were 0.997, 97.41%, 97.07% and 96.82%, respectively. For test data set (n=273 images), the AUC, sensitivity, accuracy and specificity of the DTL for correctly classification funds images were 0.926, 88.17%, 87.18% and 86.67%, respectively.ConclusionIn the evaluation, the DTL presented high sensitivity and specificity for detecting abnormal fundus-related diseases. Further research is necessary to improve this method and evaluate the applicability of the DTL in the community health care center.

Application of Dempster–Shafer Data Fusion Technique in Support of Decision Making with Big Data

2017 ◽  
Vol 2645 (1) ◽  
pp. 32-37 ◽  
Author(s):  
Ping Yi ◽  
Songling Zhang
Keyword(s):  
Decision Making ◽  
Big Data ◽  
Data Fusion ◽  
Incomplete Information ◽  
Data Sources ◽  
Fusion Technique ◽  
Crowd Management ◽  
Multiple Data ◽  
Data Source ◽  
Single Data

This paper introduces applications of the Dempster–Shafer (D-S) data fusion technique in transportation system decision making. D-S inference is a statistics-based data classification technique, and it can be used when data sources contribute discontinuous and incomplete information and no single data source can produce an overwhelmingly high probability of certainty for identifying the most probable event. The technique captures and combines the information contributed by the data sources by using Dempster’s rule to find the conjunction of the events and to determine the highest associated probability. The D-S theory is explained and its implementation described through numerical examples of a ride-hauling service and of crowd management at a subway station. Results from the applications have shown that the technique is very effective in dealing with incomplete information and multiple data sources in the era of big data.

Sign in / Sign up

Export Citation Format

Share Document