A CNN-LSTM-based Hybrid Deep Learning Model Enhanced with PSO-GA Optimization for Accurate Code Clone Detection

Code clone analysis is needed when software repositories contain multiple implementations of the same or closely related functionality. Such duplication can make later maintenance harder because a change made in one fragment may not be reflected in another. This paper presents a hybrid CNN–LSTM framework in which PSO and GA are used to refine the feature representation before classification. The workflow consists of source-code preparation, feature construction, optimization, and clone classification. CNN is used to learn local patterns from the code representation, whereas LSTM models dependencies that extend across the sequence. The optimized representation is evaluated on BigCloneBench, Google Code Jam, and a Java repository. The reported experiments show that the combined model provides higher recall, accuracy, and precision than the comparison models. The findings indicate that optimization-assisted deep learning can provide a practical route toward more reliable clone analysis in software repositories.

Authors

Institutions

Publication Details

Journal
International Journal of Electronics and Communication Engineering
Published
2026-11-12
DOI
https://doi.org/10.14445/23488549/ijece-v13i10p108
Primary Topic
Software Engineering Research
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A CNN-LSTM-based Hybrid Deep Learning Model Enhanced with PSO-GA Optimization for Accurate Code Clone Detection

Geetika, Tanvi -
International Journal of Electronics and Communication Engineering
Software Engineering Research
article

A CNN-LSTM-based Hybrid Deep Learning Model Enhanced with PSO-GA Optimization for Accurate Code Clone Detection

Geetika, Tanvi -
article en

Abstract

Code clone analysis is needed when software repositories contain multiple implementations of the same or closely related functionality. Such duplication can make later maintenance harder because a change made in one fragment may not be reflected in another. This paper presents a hybrid CNN–LSTM framework in which PSO and GA are used to refine the feature representation before classification. The workflow consists of source-code preparation, feature construction, optimization, and clone classification. CNN is used to learn local patterns from the code representation, whereas LSTM models dependencies that extend across the sequence. The optimized representation is evaluated on BigCloneBench, Google Code Jam, and a Java repository. The reported experiments show that the combined model provides higher recall, accuracy, and precision than the comparison models. The findings indicate that optimization-assisted deep learning can provide a practical route toward more reliable clone analysis in software repositories.

International Journal of Electronics and Communication EngineeringVol. 13(10)
Lovely Professional University (IN)
Software Engineering Research
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon