LE QUY DON
Technical University
VietnameseClear Cookie - decide language by browser settings

SibStCNN and TBCNN + kNN-TED: New models over tree structures for source code classification

Phan, A.V. and Nguyen, M.L. and Bui, L.T. (2017) SibStCNN and TBCNN + kNN-TED: New models over tree structures for source code classification. In: 18th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2017, 30 October 2017 through 1 November 2017.

Full text not available from this repository. (Upload)

Abstract

This paper aims to solve a software engineering problem by applying several approaches to exploit tree representations of programs. Firstly, we propose a new sibling-subtree convolutional neural network (SibStCNN), and combination models of tree-based neural networks and k-Nearest Neighbors (kNN) for source code classification. Secondly, we present a pruning tree technique to reduce data dimension and strengthen classifiers. The experiments show that the proposed models outperform other methods, and the pruning tree leads to not only a substantial reduction in execution time but also an increase in accuracy. © Springer International Publishing AG 2017.

Item Type: Conference or Workshop Item (Paper)
Divisions: Faculties > Faculty of Information Technology
Identification Number: 10.1007/978-3-319-68935-7_14
Uncontrolled Keywords: Classification (of information); Convolution; Motion compensation; Nearest neighbor search; Neural networks; Software engineering; Abstract Syntax Trees; Combination models; Convolutional neural network; Data dimensions; K nearest neighbor (KNN); K-nearest neighbors; Substantial reduction; Tree representation; Trees (mathematics)
Additional Information: Conference code: 202239. Language of original document: English.
URI: http://eprints.lqdtu.edu.vn/id/eprint/9749

Actions (login required)

View Item
View Item