Unraveling Code Clone Dynamics in Deep Learning Frameworks
Deep Learning (DL) frameworks play a critical role in advancing AI, and their rapid growth underscores the need for a comprehensive understanding of software quality and maintainability. DL frameworks, like other systems, are prone to code clones. Code clones refer to identical or highly similar source code fragments within the same project or even across different projects. Code cloning can have positive and negative implications for software development, influencing maintenance, readability, and bug propagation. While the existing studies focus on studying clones in DL-based applications, to our knowledge, no work has been done investigating clones, their evolution, and their impact on the maintenance of DL frameworks. In this article, we aim to address the knowledge gap concerning the evolutionary dimension of code clones in DL frameworks and the extent of code reuse across these frameworks. We empirically analyze code clones in nine popular DL frameworks, i.e., TensorFlow, Paddle, PyTorch, Aesara, Ray, MXNet, Keras, Jax, and BentoML, to investigate (1) the characteristics of the long-term code cloning evolution over releases in each framework, (2) the short-term, i.e., within-release, code cloning patterns and their influence on the long-term trends, and (3) the file-level code clones within the DL frameworks. Our findings reveal that DL frameworks adopt four distinct cloning trends: “Serpentine,” “Rise and Fall,” “Decreasing,” and “Stable” and that these trends present some common and distinct characteristics. For instance, bug-fixing activities persistently happen in clones irrespective of the clone evolutionary trend but occur more in the “Serpentine” trend. Moreover, the within-release level investigation demonstrates that short-term code cloning practices impact long-term cloning trends. The cross-framework code clone investigation reveals the presence of functional and architectural adaptation file-level cross-framework code clones across the nine studied frameworks. We provide insights that foster robust clone practices and collaborative maintenance in the development of DL frameworks.
Wed 15 AprDisplayed time zone: Brasilia, Distrito Federal, Brazil change
11:00 - 12:30 | Evolution 1Journal-first Papers / Research Track at Oceania VIII Chair(s): Leopoldo Teixeira Federal University of Pernambuco | ||
11:00 15mTalk | Func: Reducing the impact of Android framework evolution on malware detection Journal-first Papers Hailong Yu Harbin Institute of Technology, Harbin Institute of Technology Zhengzhou Research Institute, Tiantian Wang Harbin Institute of Technology, Lwin Khin Shar Singapore Management University, Hanmeng Li Harbin Institute of Technology, David Lo Singapore Management University Media Attached | ||
11:15 15mTalk | Unraveling Code Clone Dynamics in Deep Learning Frameworks Journal-first Papers Maram Assi Université du Québec à Montréal, Safwat Hassan University of Toronto, Ying Zou Queen's University, Kingston, Ontario | ||
11:30 15mTalk | Remediating Superfluous Re-Rendering in React ApplicationsDistinguished Paper Award Research Track Farideh Khalili , Satyajit Gokhale Amazon, Alexi Turcotte CISPA, Dale Xu Boston University, Frank Tip Northeastern University | ||
11:45 15mTalk | Minimizing Breaking Changes and Redundancy in Mitigating Technical Lag for Java Projects Research Track Rui Lu East China Normal University, Lyuye Zhang Nanyang Technological University, Kaixuan Li Nanyang Technological University, Min Zhang East China Normal University, Yixiang Chen East China Normal University Pre-print | ||
12:00 15mTalk | CRUDinfer: Automated CRUD Semantics Inference for REST APIs Through Black-box Testing Research Track Michele Pasqua University of Verona, Davide Corradini University of Luxembourg, Michele Perlotto University of Naples, Mariano Ceccato University of Verona | ||
12:15 15mTalk | Diffploit: Facilitating Cross-Version Exploit Migration for Open Source Library Vulnerabilities Research Track Zirui Chen Zhejiang University, Zhipeng Xue Zhejiang University, Jiayuan Zhou Queen's University, Xing Hu Zhejiang University, Xin Xia Zhejiang University, Xiaohu Yang Zhejiang University Pre-print Media Attached | ||