arXiv subjects

Lionel Briand

Publications and source records attributed to Lionel Briand.

At least 37 records · Page 2Linked to original sources

arXiv2026

Mutation-Guided Unit Test Generation with a Large Language Model

Unit tests play a vital role in uncovering potential faults in software. While tools like EvoSuite focus on maximizing code coverage, recent advances in large language models (LLMs) have shifted attention toward LLM-based test generation. However, code coverage metrics -- such as line and branch coverage -- remain overly emphasized in reported research, despite being weak indicators of a test suite's fault-detection capability. In contrast, mutation score offers a more reliable and stringent measure, as demonstrated in our findings where some test suites achieve 100% coverage but only 4% mutation score. Although a few studies consider mutation score, the effectiveness of LLMs in killing mutants remains underexplored. In this paper, we propose MUTGEN, a mutation-guided, LLM-based test generation approach that incorporates mutation feedback directly into the prompt. Evaluated on 204 subjects from two benchmarks, MUTGEN significantly outperforms both EvoSuite and vanilla prompt-based strategies in terms of mutation score. Furthermore, MUTGEN introduces an iterative generation mechanism that pushes the limits of LLMs in killing additional mutants. Our study also provide insights into the limitations of LLM-based generation, analyzing the reasons for live and uncovered mutants, and the impact of different mutation operators on generation effectiveness.

cs.SE↗

arXiv2026

Hallucination to Consensus: Multi-Agent LLMs for End-to-End JUnit Test Generation

Unit testing plays a critical role in ensuring software correctness. However, writing unit tests manually is labor-intensive, especially for strongly typed languages like Java, motivating the need for automated approaches. Traditional methods primarily rely on search-based or randomized algorithms to achieve high code coverage and produce regression oracles, which are derived from the program's current behavior rather than its intended functionality. Recent advances in LLMs have enabled oracle generation from natural language descriptions, aligning better with user requirements. However, existing LLM-based methods often require fine-tuning or rely on external tools such as EvoSuite for test prefix generation, making them costly or cumbersome to apply in practice. In this work, we propose CANDOR, a novel prompt engineering-based LLM framework for automated unit test generation in Java. CANDOR orchestrates multiple specialized LLM agents to collaboratively generate complete tests. To mitigate the notorious hallucinations in LLMs and improve oracle correctness, we introduce a novel strategy that engages multiple reasoning LLMs in a panel discussion and generates accurate oracles based on consensus. Additionally, to reduce the verbosity of reasoning LLMs' outputs, we propose a novel dual-LLM pipeline to produce concise and structured oracle evaluations. Our experiments show that CANDOR is comparable with EvoSuite in generating tests with high code coverage and clearly superior in terms of mutation score. Moreover, our prompt engineering-based approach CANDOR significantly outperforms the SOTA fine-tuning-based oracle generator TOGLL by at least 21.1 percentage points in oracle correctness on both correct and faulty source code. Further ablation studies confirm the critical contributions of key agents in generating high-quality tests.

cs.SE↗

arXiv2026

Classifier or Prompt: A Case Study on Legal Requirements Traceability

New regulations are introduced to ensure software development aligns with ethical concerns and protects public safety. Showing compliance requires tracing requirements to legal provisions. Requirements traceability is a key task where engineers must analyze technical requirements against target artifacts, often within limited time. Manually analyzing complex systems with hundreds of requirements is infeasible. The legal dimension adds challenges that increase effort. In this paper, we investigate two automated solutions based on language models, including large ones (LLMs). The first solution, Kashif, is a classifier that leverages sentence transformers and semantic similarity. The second solution, RICE_LRT, prompts a recent LLM based on RICE, a prompt engineering framework. Using a publicly available benchmark dataset, we empirically evaluate Kashif and compare it against seven baseline classifiers from the literature (LSI, LDA, GloVe, TraceBERT, RoBERTa, and LLaMa). Kashif can identify trace links with F2 score of 63%, outperforming the best baseline by a substantial margin of 21 percentage points (pp) in F2 score. On a newly created and more complex requirements document traced to the European general data protection regulation (GDPR), RICE_LRT outperforms Kashif and baseline prompts in the literature by achieving an average recall of 84% and F2 score of 61%, improving the F2 score by 34 pp compared to the best baseline prompt. Our results indicate that requirements traceability in legal contexts cannot be adequately addressed by techniques proposed in the literature that are not specifically designed for legal artifacts. Furthermore, we demonstrate that our engineered prompt outperforms both classifier-based approaches and baseline prompts.

cs.SE↗

arXiv2026

A Structured Approach to Safety Case Construction for AI Systems

Safety cases, structured arguments that a system is acceptably safe, are becoming central to the governance of AI systems. Yet, traditional safety-case practices from aviation or nuclear engineering rely on well-specified system boundaries, stable architectures, and known failure modes. Modern AI systems, such as generative and agentic AI, are the opposite. Their capabilities emerge unpredictably from low-level training objectives, their behaviour varies with prompts, and their risk profiles shift through fine-tuning, scaffolding, or deployment context. This study examines how safety cases are currently constructed for AI systems and why classical approaches fail to capture these dynamics. This study introduces comprehensive taxonomies for AI-specific claim types (assertion-based, constraint-based, capability-based), argument types (demonstrative, comparative, causal/explanatory, risk-based, and normative), and evidence families (empirical, mechanistic, comparative, expert-driven, formal methods, operational/field data, and model-based). It then proposes a reusable safety-case template, each of which follows a predefined structure of claims, arguments, and evidence tailored for AI systems. Each template is illustrated by end-to-end patterns that address distinctive challenges, such as evaluation without ground truth, dynamic model updates, and threshold-based risk decisions. The result is a systematic, composable, and reusable approach to constructing and maintaining safety cases that are credible, auditable, and adaptive to the evolving behaviour of generative and frontier AI systems.

cs.SE↗

arXiv2026

SWE-ABS: Adversarial Benchmark Strengthening Exposes Inflated Success Rates on Test-based Benchmark

The SWE-Bench Verified leaderboard is approaching saturation, with the top system achieving 78.80%. However, we show that this performance is inflated. Our re-evaluation reveals that one in five "solved" patches from the top-30 agents are semantically incorrect, passing only because weak test suites fail to expose their errors. We present SWE-ABS, an adversarial framework that strengthens test suites through a two-stage pipeline: (1) coverage-driven augmentation using program slicing to target untested code regions, and (2) mutation-driven adversarial testing that synthesizes plausible but incorrect patches to expose semantic blind spots. On SWE-Bench Verified (500 instances), SWE-ABS strengthens 50.2% of instances, a 25.1x improvement over prior work, and rejects 19.71% of previously passing patches. As a result, the top agent's score decreases from 78.80% to 62.20%, leading to significant leaderboard reshuffling, with the previous top-ranked agent dropping to fifth place.

cs.SE↗

arXiv2026

Reclaiming Software Engineering as the Enabling Technology for the Digital Age

Software engineering is the invisible infrastructure of the digital age. Every breakthrough in artificial intelligence, quantum computing, photonics, and cybersecurity relies on advances in software engineering, yet the field is too often treated as a supportive digital component rather than as a strategic, enabling discipline. In policy frameworks, including major European programmes, software appears primarily as a building block within other technologies, while the scientific discipline of software engineering remains largely absent. This position paper argues that the long-term sustainability, dependability, and sovereignty of digital technologies depend on investment in software engineering research. It is a call to reclaim the identity of software engineering.

cs.SE↗

arXiv2026

Drivora: A Unified and Extensible Infrastructure for Search-based Autonomous Driving Testing

Search-based testing is critical for evaluating the safety and reliability of autonomous driving systems (ADSs). However, existing approaches are often built on heterogeneous frameworks (e.g., distinct scenario spaces, simulators, and ADSs), which require considerable effort to reuse and adapt across different settings. To address these challenges, we present Drivora, a unified and extensible infrastructure for search-based ADS testing built on the widely used CARLA simulator. Drivora introduces a unified scenario definition, OpenScenario, that specifies scenarios using low-level, actionable parameters to ensure compatibility with existing methods while supporting extensibility to new testing designs (e.g., multi-autonomous-vehicle testing). On top of this, Drivora decouples the testing engine, scenario execution, and ADS integration. The testing engine leverages evolutionary computation to explore new scenarios and supports flexible customization of core components. The scenario execution can run arbitrary scenarios using a parallel execution mechanism that maximizes hardware utilization for large-scale batch simulation. For ADS integration, Drivora provides access to 12 ADSs through a unified interface, streamlining configuration and simplifying the incorporation of new ADSs. Our tools are publicly available at https://github.com/MingfeiCheng/Drivora.

cs.SE↗

arXiv2025

LLM meets ML: Data-efficient Anomaly Detection on Unstable Logs

Most log-based anomaly detectors assume logs are stable, though logs are often unstable due to software or environmental changes. Anomaly detection on unstable logs (ULAD) is therefore a more realistic, yet under-investigated challenge. Current approaches predominantly employ machine learning (ML) models, which often require extensive labeled data for training. To mitigate data insufficiency, we propose FlexLog, a novel hybrid approach for ULAD that combines ML models -- decision tree, k-nearest neighbors, and a feedforward neural network -- with a Large Language Model (Mistral) through ensemble learning. FlexLog also incorporates a cache and retrieval-augmented generation (RAG) to further enhance efficiency and effectiveness. To evaluate FlexLog, we configured four datasets for \task, namely ADFA-U, LOGEVOL-U, SynHDFS-U, and SYNEVOL-U. FlexLog outperforms all baselines by at least 1.2 percentage points (pp) in F1 score while using much less labeled data (62.87 pp reduction). When trained on the same amount of data as the baselines, FlexLog achieves up to a 13 pp increase in F1 score on ADFA-U across varying training dataset sizes. Additionally, FlexLog maintains inference time under one second per log sequence, making it suitable for most applications, except latency-sensitive systems. Further analysis reveals the positive impact of FlexLog's key components: cache, RAG and ensemble learning.

cs.SE↗

arXiv2025

DiffGAN: A Test Generation Approach for Differential Testing of Deep Neural Networks for Image Analysis

Deep Neural Networks (DNNs) are increasingly deployed across applications. However, ensuring their reliability remains a challenge, and in many situations, alternative models with similar functionality and accuracy are available. Traditional accuracy-based evaluations often fail to capture behavioral differences between models, especially with limited test datasets, making it difficult to select or combine models effectively. Differential testing addresses this by generating test inputs that expose discrepancies in DNN model behavior. However, existing approaches face significant limitations: many rely on model internals or are constrained by available seed inputs. To address these challenges, we propose DiffGAN, a black-box test image generation approach for differential testing of DNN models. DiffGAN leverages a Generative Adversarial Network (GAN) and the Non-dominated Sorting Genetic Algorithm II to generate diverse and valid triggering inputs that reveal behavioral discrepancies between models. DiffGAN employs two custom fitness functions, focusing on diversity and divergence, to guide the exploration of the GAN input space and identify discrepancies between models' outputs. By strategically searching this space, DiffGAN generates inputs with specific features that trigger differences in model behavior. DiffGAN is black-box, making it applicable in more situations. We evaluate DiffGAN on eight DNN model pairs trained on widely used image datasets. Our results show DiffGAN significantly outperforms a SOTA baseline, generating four times more triggering inputs, with greater diversity and validity, within the same budget. Additionally, the generated inputs improve the accuracy of a machine learning-based model selection mechanism, which selects the best-performing model based on input characteristics and can serve as a smart output voting mechanism when using alternative models.

cs.CV↗

arXiv2025

MetaSel: A Test Selection Approach for Fine-tuned DNN Models

Deep Neural Networks (DNNs) face challenges during deployment due to covariate shift, i.e., data distribution shifts between development and deployment contexts. Fine-tuning adapts pre-trained models to new contexts requiring smaller labeled sets. However, testing fine-tuned models under constrained labeling budgets remains a critical challenge. This paper introduces MetaSel, a new approach tailored for DNN models that have been fine-tuned to address covariate shift, to select tests from unlabeled inputs. MetaSel assumes that fine-tuned and pre-trained models share related data distributions and exhibit similar behaviors for many inputs. However, their behaviors diverge within the input subspace where fine-tuning alters decision boundaries, making those inputs more prone to misclassification. Unlike general approaches that rely solely on the DNN model and its input set, MetaSel leverages information from both the fine-tuned and pre-trained models and their behavioral differences to estimate misclassification probability for unlabeled test inputs, enabling more effective test selection. Our extensive empirical evaluation, comparing MetaSel against 11 state-of-the-art approaches and involving 68 fine-tuned models across weak, medium, and strong distribution shifts, demonstrates that MetaSel consistently delivers significant improvements in Test Relative Coverage (TRC) over existing baselines, particularly under highly constrained labeling budgets. MetaSel shows average TRC improvements of 28.46% to 56.18% over the most frequent second-best baselines while maintaining a high TRC median and low variability. Our results confirm MetaSel's practicality, robustness, and cost-effectiveness for test selection in the context of fine-tuned models.

cs.LG↗

arXiv2025

Fuzzing-based Mutation Testing of C/C++ Software in Cyber-Physical Systems

Mutation testing can help minimize the delivery of faulty software. Therefore, it is a recommended practice for developing embedded software in safety-critical cyber-physical systems (CPS). However, state-of-the-art mutation testing techniques for C and C++ software, which are common languages for CPS, depend on symbolic execution. Unfortunately, symbolic execution's limitations hinder its applicability (e.g., systems with black-box components). We propose relying on fuzz testing, which has demonstrated its effectiveness for C and C++ software. Fuzz testing tools automatically create test inputs that explore program branches in various ways, exercising statements in different program states, and thus enabling the detection of mutants, which is our objective. We empirically evaluated our approach using software components from operational satellite systems. Our assessment shows that our approach can detect between 40% and 90% of the mutants not detected by developers' test suites. Further, we empirically determined that the best results are obtained by integrating the Clang compiler, a memory address sanitizer, and relying on laf-intel instrumentation to collect coverage and guide fuzzing. Our approach detects a significantly higher percentage of live mutants compared to symbolic execution, with an increase of up to 50 percentage points; further, we observed that although the combination of fuzzing and symbolic execution leads to additional mutants being killed, the benefits are minimal (a gain of less than one percentage point).

cs.SE↗

arXiv2025

A Comprehensive Study of Machine Learning Techniques for Log-Based Anomaly Detection

Growth in system complexity increases the need for automated log analysis techniques, such as Log-based Anomaly Detection (LAD). While deep learning (DL) methods have been widely used for LAD, traditional machine learning (ML) techniques can also perform well depending on the context and dataset. Semi-supervised techniques deserve the same attention as they offer practical advantages over fully supervised methods. Current evaluations mainly focus on detection accuracy, but this alone is insufficient to determine the suitability of a technique for a given LAD task. Other aspects to consider include training and prediction times as well as the sensitivity to hyperparameter tuning, which in practice matters to engineers. This paper presents a comprehensive empirical study evaluating a wide range of supervised and semi-supervised, traditional and deep ML techniques across four criteria: detection accuracy, time performance, and sensitivity to hyperparameter tuning in both detection accuracy and time performance. The experimental results show that supervised traditional and deep ML techniques fare similarly in terms of their detection accuracy and prediction time on most of the benchmark datasets considered in our study. Moreover, overall, sensitivity analysis to hyperparameter tuning with respect to detection accuracy shows that supervised traditional ML techniques are less sensitive than deep learning techniques. Further, semi-supervised techniques yield significantly worse detection accuracy than supervised techniques.

cs.SE↗

arXiv2025

Testing CPS with Design Assumptions-Based Metamorphic Relations and Genetic Programming

Cyber-Physical Systems (CPSs) software is used to enforce desired behaviours on physical systems. To test the interaction between the CPS software and the system's physics, engineers provide traces of desired physical states and observe traces of the actual physical states. CPS requirements describe how closely the actual physical traces should track the desired traces. These requirements are typically defined for specific, simple input traces such as step or ramp sequences, and thus are not applicable to arbitrary inputs. This limits the availability of oracles for CPSs. Our recent work proposes an approach to testing CPS using control-theoretical design assumptions instead of requirements. This approach circumvents the oracle problem by leveraging the control-theoretical guarantees that are provided when the design assumptions are satisfied. To address the test case generation and oracle problems, researchers have proposed metamorphic testing, which is based on the study of relations across tests, i.e., metamorphic relations (MRs). In this work, we define MRs based on the design assumptions and explore combinations of these MRs using genetic programming to generate CPS test cases. This enables the generation of CPS input traces with potentially arbitrary shapes, together with associated expected output traces. We use the deviation from the expected output traces to guide the generation of input traces that falsify the MRs. Our experiment results show that the MR-falsification provides engineers with new information, helping them identify passed and failed test cases. Furthermore, we show that the generation of traces that falsify the MRs is a non-trivial problem, which is successfully addressed by our genetic search.

cs.SE↗

arXiv2025

Search-based DNN Testing and Retraining with GAN-enhanced Simulations

In safety-critical systems (e.g., autonomous vehicles and robots), Deep Neural Networks (DNNs) are becoming a key component for computer vision tasks, particularly semantic segmentation. Further, since the DNN behavior cannot be assessed through code inspection and analysis, test automation has become an essential activity to gain confidence in the reliability of DNNs. Unfortunately, state-of-the-art automated testing solutions largely rely on simulators, whose fidelity is always imperfect, thus affecting the validity of test results. To address such limitations, we propose to combine meta-heuristic search, used to explore the input space using simulators, with Generative Adversarial Networks (GANs), to transform the data generated by simulators into realistic input images. Such images can be used both to assess the DNN performance and to retrain the DNN more effectively. We applied our approach to a state-of-the-art DNN performing semantic segmentation and demonstrated that it outperforms a state-of-the-art GAN-based testing solution and several baselines. Specifically, it leads to the largest number of diverse images leading to the worst DNN performance. Further, the images generated with our approach, lead to the highest improvement in DNN performance when used for retraining. In conclusion, we suggest to always integrate GAN components when performing search-driven, simulator-based testing.

cs.SE↗

arXiv2025

Automated Test Case Repair Using Language Models

Ensuring the quality of software systems through testing is essential, yet maintaining test cases poses significant challenges and costs. The need for frequent updates to align with the evolving system under test often entails high complexity and cost for maintaining these test cases. Further, unrepaired broken test cases can degrade test suite quality and disrupt the software development process, wasting developers' time. To address this challenge, we present TaRGET (Test Repair GEneraTor), a novel approach leveraging pre-trained code language models for automated test case repair. TaRGET treats test repair as a language translation task, employing a two-step process to fine-tune a language model based on essential context data characterizing the test breakage. To evaluate our approach, we introduce TaRBench, a comprehensive benchmark we developed covering 45,373 broken test repairs across 59 open-source projects. Our results demonstrate TaRGET's effectiveness, achieving a 66.1% exact match accuracy. Furthermore, our study examines the effectiveness of TaRGET across different test repair scenarios. We provide a practical guide to predict situations where the generated test repairs might be less reliable. We also explore whether project-specific data is always necessary for fine-tuning and if our approach can be effective on new projects.

cs.SE↗

arXiv2024

TEASMA: A Practical Methodology for Test Adequacy Assessment of Deep Neural Networks

Successful deployment of Deep Neural Networks (DNNs) requires their validation with an adequate test set to ensure a sufficient degree of confidence in test outcomes. Although well-established test adequacy assessment techniques have been proposed for DNNs, we still need to investigate their application within a comprehensive methodology for accurately predicting the fault detection ability of test sets and thus assessing their adequacy. In this paper, we propose and evaluate TEASMA, a comprehensive and practical methodology designed to accurately assess the adequacy of test sets for DNNs. In practice, TEASMA allows engineers to decide whether they can trust high-accuracy test results and thus validate the DNN before its deployment. Based on a DNN model's training set, TEASMA provides a procedure to build accurate DNN-specific prediction models of the Fault Detection Rate (FDR) of a test set using an existing adequacy metric, thus enabling its assessment. We evaluated TEASMA with four state-of-the-art test adequacy metrics: Distance-based Surprise Coverage (DSC), Likelihood-based Surprise Coverage (LSC), Input Distribution Coverage (IDC), and Mutation Score (MS). Our extensive empirical evaluation across multiple DNN models and input sets such as ImageNet, reveals a strong linear correlation between the predicted and actual FDR values derived from MS, DSC, and IDC, with minimum R^2 values of 0.94 for MS and 0.90 for DSC and IDC. Furthermore, a low average Root Mean Square Error (RMSE) of 9% between actual and predicted FDR values across all subjects, when relying on regression analysis and MS, demonstrates the latter's superior accuracy when compared to DSC and IDC, with RMSE values of 0.17 and 0.18, respectively. Overall, these results suggest that TEASMA provides a reliable basis for confidently deciding whether to trust test results for DNN models.

cs.SE↗

arXiv2024

LTM: Scalable and Black-box Similarity-based Test Suite Minimization based on Language Models

Test suites tend to grow when software evolves, making it often infeasible to execute all test cases with the allocated testing budgets, especially for large software systems. Test suite minimization (TSM) is employed to improve the efficiency of software testing by removing redundant test cases, thus reducing testing time and resources, while maintaining the fault detection capability of the test suite. Most existing TSM approaches rely on code coverage (white-box) or model-based features, which are not always available to test engineers. Recent TSM approaches that rely only on test code (black-box) have been proposed, such as ATM and FAST-R. To address the scalability, we propose LTM (Language model-based Test suite Minimization), a novel, scalable, and black-box similarity-based TSM approach based on large language models (LLMs), which is the first application of LLMs in the context of TSM. To support similarity measurement for test code embeddings, we investigate five pre-trained language models: CodeBERT, GraphCodeBERT, UniXcoder, StarEncoder, and CodeLlama, on which we compute two similarity measures: Cosine Similarity and Euclidean Distance. Our goal is to find similarity measures that are not only computationally more efficient but can also better guide a Genetic Algorithm (GA) to search for optimal minimized test suites, thus reducing the overall search time. Experimental results show that the best configuration of LTM (UniXcoder/Cosine) outperforms ATM in three aspects: (a) achieving a slightly greater saving rate of testing time (41.72% versus 41.02%, on average); (b) attaining a significantly higher fault detection rate (0.84 versus 0.81, on average); and, most importantly, (c) minimizing test suites nearly five times faster on average, with higher gains for larger test suites and systems, thus achieving much higher scalability.

cs.SE↗

arXiv2024

FlakyFix: Using Large Language Models for Predicting Flaky Test Fix Categories and Test Code Repair

Flaky tests are problematic because they non-deterministically pass or fail for the same software version under test, causing confusion and wasting development effort. While machine learning models have been used to predict flakiness and its root causes, there is much less work on providing support to fix the problem. To address this gap, in this paper, we focus on predicting the type of fix that is required to remove flakiness and then repair the test code on that basis. We do this for a subset of flaky tests where the root cause of flakiness is in the test itself and not in the production code. One key idea is to guide the repair process with additional knowledge about the test's flakiness in the form of its predicted fix category. Thus, we first propose a framework that automatically generates labeled datasets for 13 fix categories and trains models to predict the fix category of a flaky test by analyzing the test code only. Our experimental results using code models and few-shot learning show that we can correctly predict most of the fix categories. To show the usefulness of such fix category labels for automatically repairing flakiness, we augment the prompts of GPT-3.5 Turbo, a Large Language Model (LLM), with such extra knowledge to request repair suggestions. The results show that our suggested fix category labels, complemented with in-context learning, significantly enhance the capability of GPT-3.5 Turbo in generating fixes for flaky tests. Based on the execution and analysis of a sample of GPT-repaired flaky tests, we estimate that a large percentage of such repairs (roughly between 51% and 83%) can be expected to pass. For the failing repaired tests, on average, 16% of the test code needs to be further changed for them to pass.

cs.SE↗