Unmanned aerial vehicles (UAVs) are becoming increasingly important and widely used in modern society. Software bugs these systems can cause severe issues, such as system crashes, hangs, undefined behaviors. Some also be exploited by hackers to launch security attacks, resulting catastrophic consequences. Therefore, techniques that help detect fix software UAVs highly desirable. However, although there many existing studies on various types of software, the characteristics UAV have never...

Optimizing low-protein (LP) diets in swine nutrition is critical for reducing nitrogen excretion and resource waste, while meat quality. However, LP may disrupt amino acid (AA) balance, affecting growth health. Supplementing with non-protein sources such as diammonium phosphate (DP) can enhance utilization support protein synthesis efficiently. This study aimed to investigate the effects of adding DP on performance, organ indices, carcass traits, quality, AA composition growing pigs....

This paper presents an empirical study of 5741 bug reports for the Linux kernel from evolutionary perspective, with aim obtaining a deep understanding characteristics in operating system. Bug classification is performed based on fault triggering conditions, followed by analysis proportions and evolution types as well comparisons among versions, products, repair locations. In addition, regression bugs relationship between time needed to fix them are presented. Moreover, procedure type complex...

TensorFlow is one of the most popular machine learning frameworks for developing algorithms. Because popularity and large-scale use TensorFlow, even a single bug may lead to severe consequences impact large number users. With growing safety-critical systems built upon its reliability becoming increasingly important. An essential step ensure TensorFlow's understand characteristics bugs that occurred in TensorFlow. This paper presents first comprehensive empirical study on fault triggering...

Duplicate bug reports often exist in tracking systems (BTSs). Almost all the existing approaches for automatically detecting duplicate are based on text similarity. A recent study found that such may become ineffective duplicates submitted after just-in-time (JIT) retrieval, which is now a built-in feature of modern BTSs (e.g., Bugzilla). This mainly because embedded JIT suggests possible database when reporter types new summary field, therefore minimizing submission textually similar...

Understanding and predicting types of bugs are practical importance for developers to improve the testing efficiency take appropriate steps address in software releases. However, due complex conditions under which faults manifest complexity classification rules, automatic Mandelbugs is a difficult task. In this article, we present deep semantic information-based Mandelbug method that combines model with learning classifier makes use both labeled unlabeled bug reports. By training report on...

Deep learning (DL) systems are complex component-based systems, which consist of core program (code implementation and data), Python (language interpreter), third-party libraries, low-level development tools, OS, hardware environments. Incompatible interaction between components would cause serious compatibility issues, substantially affecting the deployment processes. What types issues frequently exposed in DL systems? root causes such how do developers fix them? How far we from...

With the flourishing development of Unmanned Aerial Vehicles (UAVs), mission tasks UAVs have become more and complex. Consequently, a Real-Time Operating System (RTOS) that provides operating environments for various services on these has crucial, which leads to necessity having deep understanding an RTOS. In this paper, empirical study is conducted FreeRTOS, commonly used RTOS UAVs, from complex network perspective. A total 85 releases V2.4.2 V10.0.0, are modeled as directed networks, in...

Understanding the types of defects is practical interest, which could help developers adopt proper measures in current and future software releases. As amount bug reports increasing, manual classification brings a heavy burden to developers. In this paper, we propose word2vec based framework multi-granularity automatic for on fault triggers. Except classifying into bug/non-bug Bohrbug/Mandelbug, Mandelbugs focus paper. Characteristic representation common methods suffer from data sparsity...

Linux operating system is a complex that prone to suffer failures during usage, and increases difficulties of fixing bugs. Different testing strategies fault mitigation methods can be developed applied based on different types bugs, which leads the necessity have deep understanding nature bugs in Linux. In this paper, an empirical study carried out 5741 bug reports kernel from evolution perspective. A classification conducted triggering conditions, followed by analysis type proportions over...

Regression bugs are a type of that cause feature software worked correctly but stop working after certain commit. This paper presents systematic study regression bug chains, an important unexplored phenomenon bugs. Our is based on the observation commit c1, which fixes b1, may accidentally introduce another b2. Likewise, c2 repairing b2 b3, resulting in chain, i.e., b1 → c1 b3. We have conducted large-scale by collecting 1579 and 2630 commits from 57 Linux versions (from 2.6.12 to 4.9). The...

Traceability link recovery (TLR) is an important software engineering task for developing trustworthy and reliable systems. Recently proposed deep learning (DL) models have shown their effectiveness compared to traditional information retrieval-based methods. DL often heavily relies on sufficient labeled data train the model. However, manually labeling traceability links time-consuming, labor-intensive, requires specific knowledge from domain experts. As a result, typically only small...

With the wide deployment of deep learning (DL) systems, research in reliable and robust DL is not an option but a priority, especially for safety-critical applications. Unfortunately, systems are usually nondeterministic. Due to software-level (e.g., randomness) hardware-level GPUs or CPUs) factors, multiple training runs can generate inconsistent models yield different evaluation results, even with identical settings data on same implementation framework hardware platform. Existing studies...

For a particular type of aircraft flight simulation, how to accurately describe the control characteristic is problem. In this paper, research longitudinal law identification based on QAR (quick access recorder) data, first all, trim and linearization method B737-800 six-dof nonlinear model deduced, linear state equation typical working point in cruise phase given. Then, basis PID structure, identify phase, parameters are obtained. Finally gives simulation example, results compared with...

Unmanned aerial vehicles (UAVs) are becoming increasingly ubiquitous in our daily lives. However, like many other complex systems, UAVs susceptible to software bugs that can lead abnormal system behaviors and undesirable consequences. It is crucial study such bug-induced UAV anomalies, which often manifested flight logs, help assure the quality safety of systems. there has been limited research on investigating code-level patterns anomalies. This impedes development effective tools for...

In modern software development, Python third-party libraries have become crucial, particularly due to their widespread use in fields such as deep learning and scientific computing. However, the parameters of APIs often change during evolution, causing compatibility issues for client applications that depend on specific versions. Due Python's flexible parameter-passing mechanism, different methods parameter passing can result API compatibility. Currently, no tool is capable automatically...

The structures and behaviors of modern software systems are more complicated, thus the modeling runtime is difficult. In this paper, we model Linux operating system (LOS) as a weighted network to investigate execution process LOS. topologies LOS analyzed it found that weight distribution follows power-law distribution. For better understanding LOS, explore manifestations components. result shows component management plays key role in Moreover, an assessment reliability proposed by...

Android piggybacked malware (i.e., apps that piggyback malicious code) are becoming ubiquitous in app stores. Malware writers often use obfuscation techniques to obfuscate evade detection by detectors. Previous studies this field have focused on the impact of code obfuscations malware, but deobfuscation detecting obfuscated has rarely been studied. Knowing about can provide useful insights into and therefore design resilient In paper we conduct an empirical study deobfuscations apps,...

The availability of traceability links between issues and commits is crucial for effective software maintenance, but these are often missing. Previous research has proposed link recovery (TLR) methods, with deep learning (DL) being the state-of-the-art approach. However, training high-performance DL models requires substantial labeled data, which can be costly to obtain. To overcome this challenge, we investigate integration semi-supervised (SSL) improve performance TLR, aiming achieve...