Advancements in Intelligent Systems
A Novel Framework for Java Based Data Race Detection using ECC-EVKM-BERT and SS-LSGRU with KL-FBIS
Authors: Devesh Lowe and Mithilesh Kumar Dubey
Publishing Date: 02-01-2025
ISBN: 978-81-975670-3-2
Abstract
When two or more threads visit a shared variable concurrently and at least one of those accesses is a write operation without the necessary synchronization mechanisms in place, it’s known as a data race in multi-threaded algorithms. Unpredictable behavior, such as damaged data, software crashes, and inconsistent outputs, may result from this. Since data races can appear inconsistently based on thread scheduling and time, they are infamously hard to debug. Data races can have serious consequences, such as unstable software that responds erratically to various stimuli. They jeopardize data integrity, which can lead to subtle errors that are challenging to find and correct. Furthermore, they may give rise to security flaws since unauthorized parties may be able to take advantage of the access to data. Numerous methods, including hybrid methods, dynamic analysis, and static analysis, are used to identify data races. Deep Learning algorithms, recently, have proved to be an effective way to identify new data races more accurately. Through the identification of intricate patterns and anomalies that conventional methods might overlook, these techniques can be used to examine big codebases and execution patterns in order to understand how to spot data races. To improve the accuracy and recall of detection efforts, models that predict the likelihood of data races can be trained using labeled datasets including multi-threaded code. In this work, the authors have proposed a framework for identifying data races using Java as implementation language, Galois Chameleon Swarm Optimization Algorithms for identification of test cases, K-Means for efficient clustering process, Linear Scaling Gated Recurrent Unit for classification accuracy. The proposed model thrives to reduce false positives and false negatives in identifying data races.
Keywords
Dynamic Data Race Detection, Machine Learning, Machine Learning, Deep Learning.
Cite as
Devesh Lowe and Mithilesh Kumar Dubey, "A Novel Framework for Java Based Data Race Detection using ECC-EVKM-BERT and SS-LSGRU with KL-FBIS", In: Harish Sharma, Chetan Sharma and Vaishali Maheshwari (eds), Advancements in Intelligent Systems, SCRS, India, 2025, pp. 17-23. https://doi.org/10.56155/978-81-975670-3-2-2