A Hybrid Feature Extraction Algorithm for Devanagari Script

Author(s):  
Deepti Khanduja ◽  
Neeta Nain ◽  
Subhash Panwar
Author(s):  
Sulis Sandiwarno

In order to solve some problems of importance of words and missing relations of semantic between words in the emotional analysis of e-learning systems, the TF-IWF algorithm weighted Word2vec algorithm model was proposed as a feature extraction algorithm. Moreover, to support this study, we employ Multinomial Naïve Bayes (MNB) to obtain more accurate results. There are three mainly steps, firstly, TF-IWF is employed used to compute the weight of word. Second, Word2vec algorithm is adopted to compute the vector of words, Third, we concatenate first and second steps. Finally, the users' opinions data is trained and classified through several machine learning classifiers especially MNB classifier. The experimental results indicate that the proposed method outperformed against previous approaches in terms of precision, recall, F-Score, and accuracy.


Author(s):  
Malcolm C. Fields ◽  
D. C. Anderson

Abstract A hybrid feature extraction algorithm for extracting cavity features for machining applications is presented. The algorithm operates on both a feature-based solid model of a part and its corresponding boundary representation solid model. Information available from both part representations is used, offering a more robust and efficient solution for some of the critical limitations of current feature extraction algorithms, such as verification of completeness, computation of cavity volumes, and maintenance of design information. The hybrid feature extraction algorithm combines the strengths of feature-based design and feature extraction approaches to linking design and manufacturing. Starting with a feature-based model of a part consisting of volumetric design features combined with a stock shape using set operations, the algorithm transforms this model into a feature model containing only machinable cavity features. The transformation involves computations on both the set theoretic feature model and its corresponding boundary representation solid model, and deals with the complexities of feature-feature intersections and protrusions. By combining the higher-level product information contained in the design feature model with the topological and geometric information in the boundary representation model, the algorithm supplements traditional boundary representation extraction with non-geometric product information, enabling the verification of completeness, and significantly aiding the computation of the appropriate feature volumes.


2011 ◽  
Vol 33 (7) ◽  
pp. 1625-1631 ◽  
Author(s):  
Lin Lian ◽  
Guo-hui Li ◽  
Hai-tao Wang ◽  
hao Tian ◽  
Shu-kui Xu

2012 ◽  
Vol 19 (10) ◽  
pp. 639-642 ◽  
Author(s):  
Qianwei Zhou ◽  
Guanjun Tong ◽  
Dongfeng Xie ◽  
Baoqing Li ◽  
Xiaobing Yuan

Sign in / Sign up

Export Citation Format

Share Document