arXiv subjects

Nafiseh Kahani

Publications and source records attributed to Nafiseh Kahani.

14 recordsLinked to original sources

arXiv2026

Complexity-Aware Evaluation of LLM Comprehension

Large language models (LLMs) are increasingly used for software engineering tasks that require understanding existing source code, including behavior prediction, function explanation, debugging, and code review. However, aggregate benchmark accuracy can conceal how model reliability changes as source code becomes structurally more complex. This paper presents a complexity-aware framework for evaluating LLM code comprehension using cyclomatic complexity, nesting depth, branching factor, and Halstead volume. We evaluate DeepSeek-Coder-V2 and Llama through two complementary tasks: automatic input-output prediction over 300 Python functions and manually assessed semantic comprehension over a balanced subset of 60 functions. The functions are grouped into Low-, Medium-, and High-complexity bands. DeepSeek-Coder-V2 achieves an overall automatic accuracy of 78.33%, compared with 70.33% for Llama. However, accuracy decreases substantially from Low to High complexity, from 93.52% to 52.78% for DeepSeek-Coder-V2 and from 87.04% to 47.22% for Llama. Incorrect predictions are consistently associated with higher values of all four complexity metrics, and correlation and logistic-regression analyses confirm broadly comparable negative associations between structural complexity and correctness. Manual semantic comprehension shows the same degradation pattern, with accuracy decreasing from 100.00% to 75.00% for DeepSeek-Coder-V2 and from 90.00% to 60.00% for Llama. These findings demonstrate that complexity-aware evaluation provides a more diagnostic assessment of LLM code-comprehension reliability than aggregate accuracy alone.

cs.SE↗

arXiv2026

Image Augmentation as Test Generation for Deep Learning-Based Image Retrieval Systems

Ensuring the reliability of deep learning-based image retrieval systems is a software engineering challenge. This paper presents a dual contribution: (1) a literature review of augmentation and generation techniques which resulted in the identification of 50 techniques which we organized into a ten-category taxonomy, and (2) a large-scale empirical study that evaluates these techniques as test generators for embedding-based image retrieval systems. Augmented images are embedded using Amazon Titan and OpenCLIP, and evaluated across four analytical dimensions: (1) embedding-space similarity, (2) embedding uncertainty measured via four estimators, (3) semantic realism scored by LLaVA, and (4) retrieval failure rate. Experiments are performed on three datasets: CIFAR-10, ImageNet-1K, and a dataset from an industrial partner (March Networks). Across all evaluated datasets and embedding models, and under the single severity level tested for each technique, weather simulation and SaSPA are the image augmentation/generation techniques that produce the highest embedding uncertainty and failure rates while maintaining a favorable balance between performance stability, visual realism, and augmentation effectiveness. The results we discuss are configuration-specific and may shift under milder or stronger perturbation settings. In contrast, GAN-based augmentation techniques are among the lowest in realism, indicating the presence of synthetic artifacts and perceptual inconsistencies that reduce their suitability to produce realistic test inputs. Overall, our findings provide practical guidelines for selecting augmentation techniques that maximize test diversity while preserving realistic image characteristics, thereby enabling the construction of comprehensive and effective test suites for image retrieval systems while reducing the cost of manual data labeling through the use of metamorphic testing.

cs.SE↗

arXiv2026

Runtime Compliance Verification for AI Agents

AI agents now handle personal data through tool use, function calls, and multi turn dialogue, which can create obligations under the General Data Protection Regulation (GDPR). Current testing practices mainly rely on offline red teaming or static prompt review, but they do not guarantee at runtime that agent behavior follows regulatory rules. We propose C-Trace (Compliance Trace based Runtime Agent Conformance Enforcement), a verification framework that: (i) expresses a subset of GDPR requirements, including consent, purpose limitation, data minimization, and the right to erasure, as formal policy predicates over agent execution traces; (ii) uses a runtime monitor that intercepts every tool invocation and model output and rejects non-compliant actions; and (iii) tests the agent with attack dialogues, including DSPy generated prompts and verbatim prompts from red teaming corpora, that try to induce violations. We evaluate the framework on four case studies reframed to GDPR. Under 10 percent per-category extractor noise, including drop-out and over-typing, the monitor keeps the attack success rate at less than or equal to 12 percent, below the baselines we compare against, and false positives at less than or equal to 16 percent, and reaches 0 percent ASR under perfect extraction.

cs.SE↗

arXiv2026

Teaching Machine Learning to Software Engineers

Machine learning (ML) and Artificial Intelligence (AI) components are increasingly embedded in software products, yet undergraduate software engineering (SE) curricula rarely provide systematic preparation for building, testing, deploying, and maintaining AI/ML-based software systems. This paper aims to provide evidence-based guidance for integrating AI/MLrelevant content into core SE education. We compile and define a structured inventory of topics relevant to SE practice in AI/MLbased software, then map these topics against required courses in a set of representative SE curricula to identify coverage gaps. To assess educational priorities and feasibility, we survey SE instructors on topic importance and integration constraints. Based on the crosswalk between topic definitions, curriculum coverage, and instructor prioritization, we derive a guideline that recommends where and how high-priority topics can be embedded within existing SE courses.

cs.SE↗

arXiv2026

Testing Agentic Workflows with Structural Coverage Criteria

Multi-agent systems increasingly expose explicit workflow structure: agents, tools, tool-access rules, restrictions, and delegation paths. Existing evaluations rely largely on end-to-end task success, benchmark scores, final-response quality, or prompt-level checks, which provide limited evidence that this declared coordination structure has actually been exercised. This makes it difficult to assess test-suite adequacy or detect structural regressions in tool access, restrictions, and inter-agent delegation. We address this gap with a structural testing approach for multi-agent workflow specifications. The approach represents each workflow as a typed coordination graph, derives coverage obligations over reachable agents, allowed tool edges, restricted tool edges, and delegation edges, and uses coverage-driven generation with DSPy-based scenario realization to produce executable tests. The graph fixes what must be covered; DSPy realizes those obligations as natural-language scenarios whose witnesses are checked at runtime. We implement the approach for OpenAI Agents SDK-style workflows and evaluate it on ten SDK-derived benchmarks comprising 49 reachable agents, 47 tools, and 403 structural obligations. Generated scenarios witness 54/75 allowed-tool obligations and 36/48 delegation obligations within a bounded refinement budget. The adversarial restricted-tool criterion elicits 23/248 restricted-call violations, separating workflows whose restrictions hold under probing from workflows with concrete misrouting failures. These results show that structural coverage provides a useful adequacy layer for multi-agent workflow testing: it does not replace semantic or end-to-end evaluation, but reveals whether declared agents, tool-access rules, restrictions, and delegation paths have been exercised.

cs.SE↗

arXiv2026

Prompt Optimization for LLM Code Generation via Reinforcement Learning

Large Language Models (LLMs) can generate code from natural language, but their performance is highly sensitive to prompt formulation. We propose a reinforcement-learning-based framework that models prompt refinement as a sequential decision-making problem. A Proximal Policy Optimization (PPO) agent iteratively improves prompts using a hybrid action space that combines direct generation, genetic lexical mutation and semantic rewriting, guided by shaped rewards derived from unit-test feedback. We evaluate the framework on MBPP+, HumanEval+, and APPS using CodeT5+, CodeLLaMA, and DeepSeek-Coder as frozen code generators. On the 500-task MBPP+ test set, the PPO agent achieves strict Pass@1 scores of 57.58%, 64.80%, and 85.50%, respectively, outperforming EPiC, Reflexion, and Random-Hybrid. Soft-Pass@1 reaches 67.90%, 73.10%, and 88.20%, respectively. Similar improvements are observed on HumanEval+ and APPS across all backbone models. The results demonstrate that reinforcement learning with shaped test-driven rewards improves functional correctness in LLM-based code generation.

cs.SE↗

arXiv2026

A Privacy by Design Framework for Large Language Model-Based Applications for Children

Children are increasingly using technologies powered by Artificial Intelligence (AI). However, there are growing concerns about privacy risks, particularly for children. Although existing privacy regulations require companies and organizations to implement protections, doing so can be challenging in practice. To address this challenge, this article proposes a framework based on Privacy-by-Design (PbD), which guides designers and developers to take on a proactive and risk-averse approach to technology design. Our framework includes principles from several privacy regulations, such as the General Data Protection Regulation (GDPR) from the European Union, the Personal Information Protection and Electronic Documents Act (PIPEDA) from Canada, and the Children's Online Privacy Protection Act (COPPA) from the United States. We map these principles to various stages of applications that use Large Language Models (LLMs), including data collection, model training, operational monitoring, and ongoing validation. For each stage, we discuss the operational controls found in the recent academic literature to help AI service providers and developers reduce privacy risks while meeting legal standards. In addition, the framework includes design guidelines for children, drawing from the United Nations Convention on the Rights of the Child (UNCRC), the UK's Age-Appropriate Design Code (AADC), and recent academic research. To demonstrate how this framework can be applied in practice, we present a case study of an LLM-based educational tutor for children under 13. Through our analysis and the case study, we show that by using data protection strategies such as technical and organizational controls and making age-appropriate design decisions throughout the LLM life cycle, we can support the development of AI applications for children that provide privacy protections and comply with legal requirements.

cs.AI↗

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

Understanding Defects in Generated Codes by Language Models

This study investigates the reliability of code generation by Large Language Models (LLMs), focusing on identifying and analyzing defects in the generated code. Despite the advanced capabilities of LLMs in automating code generation, ensuring the accuracy and functionality of the output remains a significant challenge. By using a structured defect classification method to understand their nature and origins this study categorizes and analyzes 367 identified defects from code snippets generated by LLMs, with a significant proportion being functionality and algorithm errors. These error categories indicate key areas where LLMs frequently fail, underscoring the need for targeted improvements. To enhance the accuracy of code generation, this paper implemented five prompt engineering techniques, including Scratchpad Prompting, Program of Thoughts Prompting, Chain-of-Thought Prompting, Chain of Code Prompting, and Structured Chain-of-Thought Prompting. These techniques were applied to refine the input prompts, aiming to reduce ambiguities and improve the models' accuracy rate. The research findings suggest that precise and structured prompting significantly mitigates common defects, thereby increasing the reliability of LLM-generated code.

cs.SE↗

arXiv2024

Code Linting using Language Models

Code linters play a crucial role in developing high-quality software systems by detecting potential problems (e.g., memory leaks) in the source code of systems. Despite their benefits, code linters are often language-specific, focused on certain types of issues, and prone to false positives in the interest of speed. This paper investigates whether large language models can be used to develop a more versatile code linter. Such a linter is expected to be language-independent, cover a variety of issue types, and maintain high speed. To achieve this, we collected a large dataset of code snippets and their associated issues. We then selected a language model and trained two classifiers based on the collected datasets. The first is a binary classifier that detects if the code has issues, and the second is a multi-label classifier that identifies the types of issues. Through extensive experimental studies, we demonstrated that the developed large language model-based linter can achieve an accuracy of 84.9% for the binary classifier and 83.6% for the multi-label classifier.

cs.SE↗

arXiv2024

An Accurate and Low-Parameter Machine Learning Architecture for Next Location Prediction

Next location prediction is a discipline that involves predicting a users next location. Its applications include resource allocation, quality of service, energy efficiency, and traffic management. This paper proposes an energy-efficient, small, and low parameter machine learning (ML) architecture for accurate next location prediction, deployable on modest base stations and edge devices. To accomplish this we ran a hundred hyperparameter experiments on the full human mobility patterns of an entire city, to determine an exact ML architecture that reached a plateau of accuracy with the least amount of model parameters. We successfully achieved a reduction in the number of model parameters within published ML architectures from 202 million down to 2 million. This reduced the total size of the model parameters from 791 MB down to 8 MB. Additionally, this decreased the training time by a factor of four, the amount of graphics processing unit (GPU) memory needed for training by a factor of twenty, and the overall accuracy was increased from 80.16% to 82.54%. This improvement allows for modest base stations and edge devices which do not have a large amount of memory or storage, to deploy and utilize the proposed ML architecture for next location prediction.

cs.LG↗

arXiv2022

Scalable and Accurate Test Case Prioritization in Continuous Integration Contexts

Continuous Integration (CI) requires efficient regression testing to ensure software quality without significantly delaying its CI builds. This warrants the need for techniques to reduce regression testing time, such as Test Case Prioritization (TCP) techniques that prioritize the execution of test cases to detect faults as early as possible. Many recent TCP studies employ various Machine Learning (ML) techniques to deal with the dynamic and complex nature of CI. However, most of them use a limited number of features for training ML models and evaluate the models on subjects for which the application of TCP makes little practical sense, due to their small regression testing time and low number of failed builds. In this work, we first define, at a conceptual level, a data model that captures data sources and their relations in a typical CI environment. Second, based on this data model, we define a comprehensive set of features that covers all features previously used by related studies. Third, we develop methods and tools to collect the defined features for 25 open-source software systems with enough failed builds and whose regression testing takes at least five minutes. Fourth, relying on the collected dataset containing a comprehensive feature set, we answer four research questions concerning data collection time, the effectiveness of ML-based TCP, the impact of the features on effectiveness, the decay of ML-based TCP models over time, and the trade-off between data collection time and the effectiveness of ML-based TCP techniques.

cs.SE↗

arXiv2021

Execution of Partial State Machine Models

The iterative and incremental nature of software development using models typically makes a model of a system incomplete (i.e., partial) until a more advanced and complete stage of development is reached. Existing model execution approaches (interpretation of models or code generation) do not support the execution of partial models. Supporting the execution of partial models at the early stages of software development allows early detection of defects, which can be fixed more easily and at a lower cost. This paper proposes a conceptual framework for the execution of partial models, which consists of three steps: static analysis, automatic refinement, and input-driven execution. First, a static analysis that respects the execution semantics of models is applied to detect problematic elements of models that cause problems for the execution. Second, using model transformation techniques, the models are refined automatically, mainly by adding decision points where missing information can be supplied. Third, refined models are executed, and when the execution reaches the decision points, it uses inputs obtained either interactively or by a script that captures how to deal with partial elements. We created an execution engine called PMExec for the execution of partial models of UML-RT (i.e., a modeling language for the development of soft real-time systems) that embodies our proposed framework. We evaluated PMExec based on several use-cases that show that the static analysis, refinement, and application of user input can be carried out with reasonable performance and that the overhead of approach, which is mostly due to the refinement and the increase in model complexity it causes, is manageable. We also discuss the properties of the refinement formally and show how the refinement preserves the original behaviors of the model.

cs.SE↗

arXiv2021

Reinforcement Learning for Test Case Prioritization

Continuous Integration (CI) significantly reduces integration problems, speeds up development time, and shortens release time. However, it also introduces new challenges for quality assurance activities, including regression testing, which is the focus of this work. Though various approaches for test case prioritization have shown to be very promising in the context of regression testing, specific techniques must be designed to deal with the dynamic nature and timing constraints of CI. Recently, Reinforcement Learning (RL) has shown great potential in various challenging scenarios that require continuous adaptation, such as game playing, real-time ads bidding, and recommender systems. Inspired by this line of work and building on initial efforts in supporting test case prioritization with RL techniques, we perform here a comprehensive investigation of RL-based test case prioritization in a CI context. To this end, taking test case prioritization as a ranking problem, we model the sequential interactions between the CI environment and a test case prioritization agent as an RL problem, using three alternative ranking models. We then rely on carefully selected and tailored state-of-the-art RL techniques to automatically and continuously learn a test case prioritization strategy, whose objective is to be as close as possible to the optimal one. Our extensive experimental analysis shows that the best RL solutions provide a significant accuracy improvement over previous RL-based work, with prioritization strategies getting close to being optimal, thus paving the way for using RL to prioritize test cases in a CI context.

cs.SE↗