arXiv subjects

Yonghui Liu

Publications and source records attributed to Yonghui Liu.

16 recordsLinked to original sources

arXiv2026

Harbor Adapters and Harbor-Index: Infrastructure and a Curated Meta-Dataset for Large-Scale Agentic Evaluation

Evaluating agents on the growing number of agentic benchmarks is challenging because they often require complex environments and agent integrations. We introduce Harbor Adapters, a unified evaluation infrastructure for agentic benchmarks. Our work makes three contributions. First, we develop benchmark adapters that port more than 80 benchmarks to evaluate arbitrary agents, and validate them through rigorous code review and parity experiments. Second, we conduct a large-scale evaluation of 8 models spanning capability tiers across 54 benchmarks; every model is run with Terminus-2 and with one of 3 native harnesses. This enables a broader analysis of agent capabilities and failure modes than was previously possible. Third, we introduce Harbor-Index, a curated set of 82 difficult, diverse, and high-quality tasks spanning 29 benchmarks, refined from the adapted suite through difficulty filtering, AI and human audit, and an audit-and-fix loop. Harbor-Index preserves the challenge and breadth of large-scale agentic evaluations while being affordable to run; no evaluated model-harness configuration exceeds 30% pass rate, and the strongest (GPT-5.5 with Codex) reaches 28.0%. We release the adapters, evaluation results, in-depth analysis, and Harbor-Index as open-source artifacts to support more reliable and comprehensive evaluation of language-model agents.

cs.AI↗

arXiv2026

Requirements-Augmented Generation for Trustworthy Acceptance Testing of LLM-Based Software

LLM-based software (LBS) integrates large language models as core components to deliver flexible, personalised responses. Unlike traditional software with deterministic outputs, LBSs exhibit context-dependent, stochastic behaviour that renders classical acceptance testing and test oracles insufficient: the same query may require fundamentally different responses depending on user personas and software context. This gap creates an urgent need for automated acceptance testing frameworks that autonomously interpret user instructions, while reliably inferring user intentions in a changing environment. In this paper, we present an automated acceptance testing framework for LBS with calibrated verdict reliability via two technical contributions. First, we introduce Requirements-Augmented Generation (REAG), which interprets user intentions by retrieving relevant software requirements, domain knowledge, and personas via adaptive RAG and self-reasoning to generate context-aware test oracles. Second, recognising that oracle generation may retrieve irrelevant constraints, misinterpret intent, or hallucinate requirements, we introduce a confidence-calibrated cascade judgment. This method quantifies verdict reliability via simulated expert agreement -- accepting high-confidence verdicts, escalating ambiguous cases, or abstaining when uncertain -- with empirical reliability guarantees backed by conformal risk control. An industrial case study on a production nutrition advisory application demonstrates that REAG achieves a 3.91/5 oracle quality score, reaching qualified or marginal oracle quality in 82% of cases. The confidence-calibrated cascade achieves 98.8% accuracy, improves oracle quality from 3.91 to 4.30 by filtering unqualified outputs, and delivers a 31.7% cost-efficiency improvement over single-judge baselines, validating industrial viability

cs.SE↗

arXiv2026

UGD: An Unsupervised Geometric Distance for Evaluating Real-world Noisy Point Cloud Denoising

Point cloud denoising is a fundamental and crucial challenge in real-world point cloud applications. Existing quantitative evaluation metrics for point cloud denoising methods are implemented in a supervised manner, which requires both the denoised point cloud and the corresponding ground-truth clean point cloud to compute a representative geometric distance. This requirement is highly problematic in real-world scenarios, where ground-truth clean point clouds are often unavailable. In this paper, we propose a simple yet effective unsupervised geometric distance (UGD) for real-world noisy point cloud denoising, calculated solely from noisy point clouds. The core idea of UGD is to learn a patch-wise prior model from a set of clean point clouds and then employ this prior model as the ground-truth to quantify the degradation by measuring the geometric variations of the denoised point cloud. To this end, we first learn a pristine Gaussian Mixture Model (GMM) with extracted patch-wise quality-aware features from a set of pristine clean point clouds by a patch-wise feature extraction network, which serves as the ground-truth for the quantitative evaluation. Then, the UGD is defined as the weighted sum of distances between each patch of the denoised point cloud and the learned pristine GMM model in the patch space. To train the employed patch-wise feature extraction network, we propose a self-supervised training framework through multi-task learning, which includes pair-wise quality ranking, distortion classification, and distortion distribution prediction. Quantitative experiments with synthetic noise confirm that the proposed UGD achieves comparable performance to supervised full-reference metrics. Moreover, experimental results on real-world data demonstrate that the proposed UGD enables unsupervised evaluation of point cloud denoising methods based exclusively on noisy point clouds.

cs.CV↗

arXiv2026

UPDA: Unsupervised Progressive Domain Adaptation for No-Reference Point Cloud Quality Assessment

While no-reference point cloud quality assessment (NR-PCQA) approaches have achieved significant progress over the past decade, their performance often degrades substantially when a distribution gap exists between the training (source domain) and testing (target domain) data. However, to date, limited attention has been paid to transferring NR-PCQA models across domains. To address this challenge, we propose the first unsupervised progressive domain adaptation (UPDA) framework for NR-PCQA, which introduces a two-stage coarse-to-fine alignment paradigm to address domain shifts. At the coarse-grained stage, a discrepancy-aware coarse-grained alignment method is designed to capture relative quality relationships between cross-domain samples through a novel quality-discrepancy-aware hybrid loss, circumventing the challenges of direct absolute feature alignment. At the fine-grained stage, a perception fusion fine-grained alignment approach with symmetric feature fusion is developed to identify domain-invariant features, while a conditional discriminator selectively enhances the transfer of quality-relevant features. Extensive experiments demonstrate that the proposed UPDA effectively enhances the performance of NR-PCQA methods in cross-domain scenarios, validating its practical applicability. The code is available at https://github.com/yokeno1/UPDA-main.

eess.IV↗

arXiv2025

CLAIM: Camera-LiDAR Alignment with Intensity and Monodepth

In this paper, we unleash the potential of the powerful monodepth model in camera-LiDAR calibration and propose CLAIM, a novel method of aligning data from the camera and LiDAR. Given the initial guess and pairs of images and LiDAR point clouds, CLAIM utilizes a coarse-to-fine searching method to find the optimal transformation minimizing a patched Pearson correlation-based structure loss and a mutual information-based texture loss. These two losses serve as good metrics for camera-LiDAR alignment results and require no complicated steps of data processing, feature extraction, or feature matching like most methods, rendering our method simple and adaptive to most scenes. We validate CLAIM on public KITTI, Waymo, and MIAS-LCEC datasets, and the experimental results demonstrate its superior performance compared with the state-of-the-art methods. The code is available at https://github.com/Tompson11/claim.

cs.RO↗

arXiv2025

Bamboo: LLM-Driven Discovery of API-Permission Mappings in the Android Framework

The permission mechanism in the Android Framework is integral to safeguarding the privacy of users by managing users' and processes' access to sensitive resources and operations. As such, developers need to be equipped with an in-depth understanding of API permissions to build robust Android apps. Unfortunately, the official API documentation by Android chronically suffers from imprecision and incompleteness, causing developers to spend significant effort to accurately discern necessary permissions. This potentially leads to incorrect permission declarations in Android app development, potentially resulting in security violations and app failures. Recent efforts in improving permission specification primarily leverage static and dynamic code analyses to uncover API-permission mappings within the Android framework. Yet, these methodologies encounter substantial shortcomings, including poor adaptability to Android SDK and Framework updates, restricted code coverage, and a propensity to overlook essential API-permission mappings in intricate codebases. This paper introduces a pioneering approach utilizing large language models (LLMs) for a systematic examination of API-permission mappings. In addition to employing LLMs, we integrate a dual-role prompting strategy and an API-driven code generation approach into our mapping discovery pipeline, resulting in the development of the corresponding tool, \tool{}. We formulate three research questions to evaluate the efficacy of \tool{} against state-of-the-art baselines, assess the completeness of official SDK documentation, and analyze the evolution of permission-required APIs across different SDK releases. Our experimental results reveal that \tool{} identifies 2,234, 3,552, and 4,576 API-permission mappings in Android versions 6, 7, and 10 respectively, substantially outprforming existing baselines.

cs.SE↗

arXiv2025

Multi-Modal Requirements Data-based Acceptance Criteria Generation using LLMs

Acceptance criteria (ACs) play a critical role in software development by clearly defining the conditions under which a software feature satisfies stakeholder expectations. However, manually creating accurate, comprehensive, and unambiguous acceptance criteria is challenging, particularly in user interface-intensive applications, due to the reliance on domain-specific knowledge and visual context that is not always captured by textual requirements alone. To address these challenges, we propose RAGcceptance M2RE, a novel approach that leverages Retrieval-Augmented Generation (RAG) to generate acceptance criteria from multi-modal requirements data, including both textual documentation and visual UI information. We systematically evaluated our approach in an industrial case study involving an education-focused software system used by approximately 100,000 users. The results indicate that integrating multi-modal information significantly enhances the relevance, correctness, and comprehensibility of the generated ACs. Moreover, practitioner evaluations confirm that our approach effectively reduces manual effort, captures nuanced stakeholder intent, and provides valuable criteria that domain experts may overlook, demonstrating practical utility and significant potential for industry adoption. This research underscores the potential of multi-modal RAG techniques in streamlining software validation processes and improving development efficiency. We also make our implementation and a dataset available.

cs.SE↗

arXiv2024

In-Context Code-Text Learning for Bimodal Software Engineering

Bimodal software analysis initially appeared to be within reach with the advent of large language models. Unfortunately, the complex interplay of natural language text and code in software engineering, presents unique challenges that prevent pretrained models to generalize to a variety of tasks. We postulate that in-context learning for the code-text bimodality is a promising avenue. This paper thus introduces a comprehensive study of in-context code-text learning, focusing on leveraging pretrained CodeLLAMA models. We consider a diverse dataset encompassing 23 software engineering tasks, which we transform in an in-context learning format. To effectively extract informative features, we propose a configurable prompt template. Our proposed pipeline, InCTRL, then unifies prompt learning across various software engineering tasks. Extensive evaluation on the study datasets demonstrates the superiority of INCTRL-models in few-shot performance, surpassing state-of-the-art models including the support model, CodeLLAMA. Typically, we observe that applied to the CodeLLAMA model, INCTRL brings improvements in terms of precision (at least about 12\%) and recall (up to 93.88\%) on various tasks. For example, on the task of program repair, INCTRL improves the BLEU score of CodeLLAMA by 85 points, while for clone detection, INCTRL achieves an improvement of 69 percentage points. Moreover, INCTRL-models offer state-of-the-art performance when using retrieval-augmented generation on individual downstream tasks. Finally, we qualitatively analyze the benefits of INCTRL over CodeLLAMA and open-source all models for broader impact. We make our code and dataset publicly available at: \begin{center} {\url{https://anonymous.4open.science/r/inctrl-B65B}} \end{center}

cs.SE↗

arXiv2024

LLM for Mobile: An Initial Roadmap

When mobile meets LLMs, mobile app users deserve to have more intelligent usage experiences. For this to happen, we argue that there is a strong need to appl LLMs for the mobile ecosystem. We therefore provide a research roadmap for guiding our fellow researchers to achieve that as a whole. In this roadmap, we sum up six directions that we believe are urgently required for research to enable native intelligence in mobile devices. In each direction, we further summarize the current research progress and the gaps that still need to be filled by our fellow researchers.

cs.SE↗

arXiv2023

ReuNify: A Step Towards Whole Program Analysis for React Native Android Apps

React Native is a widely-used open-source framework that facilitates the development of cross-platform mobile apps. The framework enables JavaScript code to interact with native-side code, such as Objective-C/Swift for iOS and Java/Kotlin for Android, via a communication mechanism provided by React Native. However, previous research and tools have overlooked this mechanism, resulting in incomplete analysis of React Native app code. To address this limitation, we have developed REUNIFY, a prototype tool that integrates the JavaScript and native-side code of React Native apps into an intermediate language that can be processed by the Soot static analysis framework. By doing so, REUNIFY enables the generation of a comprehensive model of the app's behavior. Our evaluation indicates that, by leveraging REUNIFY, the Soot-based framework can improve its coverage of static analysis for the 1,007 most popular React Native Android apps, augmenting the number of lines of Jimple code by 70%. Additionally, we observed an average increase of 84% in new nodes reached in the callgraph for these apps, after integrating REUNIFY. When REUNIFY is used for taint flow analysis, an average of two additional privacy leaks were identified. Overall, our results demonstrate that REUNIFY significantly enhances the Soot-based framework's capability to analyze React Native Android apps.

cs.SE↗

arXiv2023

Taming Android Fragmentation through Lightweight Crowdsourced Testing

Android fragmentation refers to the overwhelming diversity of Android devices and OS versions. These lead to the impossibility of testing an app on every supported device, leaving a number of compatibility bugs scattered in the community and thereby resulting in poor user experiences. To mitigate this, our fellow researchers have designed various works to automatically detect such compatibility issues. However, the current state-of-the-art tools can only be used to detect specific kinds of compatibility issues (i.e., compatibility issues caused by API signature evolution), i.e., many other essential types of compatibility issues are still unrevealed. For example, customized OS versions on real devices and semantic changes of OS could lead to serious compatibility issues, which are non-trivial to be detected statically. To this end, we propose a novel, lightweight, crowdsourced testing approach, LAZYCOW, to fill this research gap and enable the possibility of taming Android fragmentation through crowdsourced efforts. Specifically, crowdsourced testing is an emerging alternative to conventional mobile testing mechanisms that allow developers to test their products on real devices to pinpoint platform-specific issues. Experimental results on thousands of test cases on real-world Android devices show that LAZYCOW is effective in automatically identifying and verifying API-induced compatibility issues. Also, after investigating the user experience through qualitative metrics, users' satisfaction provides strong evidence that LAZYCOW is useful and welcome in practice.

cs.SE↗

arXiv2023

On the Reliability and Explainability of Language Models for Program Generation

Recent studies have adopted pre-trained language models, such as CodeT5 and CodeGPT, for automated program generation tasks like code generation, repair, and translation. Numerous language model-based approaches have been proposed and evaluated on various benchmark datasets, demonstrating promising performance. However, there is still uncertainty about the reliability of these models, particularly their realistic ability to consistently transform code sequences. This raises the question: are these techniques sufficiently trustworthy for automated program generation? Consequently, Further research is needed to understand model logic and assess reliability and explainability. To bridge these research gaps, we conduct a thorough empirical study of eight popular language models on five representative datasets to determine the capabilities and limitations of automated program generation approaches. We further employ advanced explainable AI approaches to highlight the tokens that significantly contribute to the code transformation. We discover that state-of-the-art approaches suffer from inappropriate performance evaluation stemming from severe data duplication, causing over-optimistic results. Our explainability analysis reveals that, in various experimental scenarios, language models can recognize code grammar and structural information, but they exhibit limited robustness to changes in input sequences. Overall, more rigorous evaluation approaches and benchmarks are critical to enhance the reliability and explainability of automated program generation moving forward. Our findings provide important guidelines for this goal.

cs.SE↗

arXiv2022

A Comparative Study of Smartphone and Smart TV Apps

Context: Smart TVs have become one of the most popular television types. Many app developers and service providers have designed TV versions for their smartphone applications. Despite the extensive studies on mobile app analysis, its TV equivalents receive far too little attention. The relationship between phone and TV has not been the subject of research works. Objective: In this paper, we aim to characterize the relationship between smartphone and smart TV apps. To fill this gap, we conduct a comparative study on smartphone and smart TV apps in this work, which is the starting and fundamental step to uncover the domain-specific challenges. Method: We gather a large-scale phone/TV app pairs from Google Play Store. We then analyzed the app pairs quantitatively and qualitatively from a variety of perspectives, including non-code (e.g., metadata, resources, permissions, etc.), code (e.g., components, methods, user interactions, etc.), security and privacy (e.g., reports of AndroBugs and FlowDroid). Results: Our experimental results indicate that (1) the code of the smartphone and TV apps can be released in the same app package or in separate app packages with the same package name; (2) 43% of resource files and 50% of code methods are reused between phone/TV app pairs; (3) TV and phone versions of the same app often encounter different kinds of security vulnerabilities; and (4) TV apps encounter fewer user interactions than their phone versions, but the type of user interaction events, surprisingly, are similar between phone/TV apps. Conclution: Our findings are valuable for developers and academics in comprehending the TV app ecosystem by providing additional insight into the migration of phone apps to TVs and the design mechanism of analysis tools for TV apps.

cs.SE↗

arXiv2022

A First Look at CI/CD Adoptions in Open-Source Android Apps

Continuous Integration (CI) and Continuous Delivery (CD) have been demonstrated to be effective in facilitating software building, testing, and deployment. Many research studies have investigated and subsequently improved their working processes. Unfortunately, such research efforts have largely not touched on the usage of CI/CD in the development of Android apps. We fill this gap by conducting an exploratory study of CI/CD adoption in open-source Android apps. We start by collecting a set of 84,475 open-source Android apps from the most popular three online code hosting sites, namely Github, GitLab, and Bitbucket. We then look into those apps and find that (1) only around 10\% of apps have leveraged CI/CD services, i.e., the majority of open-source Android apps are developed without accessing CI/CD services, (2) a small number of apps (291) has even adopted multiple CI/CD services, (3) nearly half of the apps adopted CI/CD services have not really used them, and (4) CI/CD services are useful to improve the popularity of projects.

cs.SE↗

arXiv2022

Automatically Recommend Code Updates: Are We There Yet?

In recent years, large pre-trained Language Models of Code (CodeLMs) have shown promising results on various software engineering tasks. One such task is automatic code update recommendation, which transforms outdated code snippets into their approved and revised counterparts. Although many CodeLM-based approaches have been proposed, claiming high accuracy, their effectiveness and reliability on real-world code update tasks remain questionable. In this paper, we present the first extensive evaluation of state-of-the-art CodeLMs for automatically recommending code updates. We assess their performance on two diverse datasets of paired updated methods, considering factors such as temporal evolution, project specificity, method size, and update complexity. Our results reveal that while CodeLMs perform well in settings that ignore temporal information, they struggle in more realistic time-wise scenarios and generalize poorly to new projects. Furthermore, CodeLM performance decreases significantly for larger methods and more complex updates. Furthermore, we observe that many CodeLM-generated "updates" are actually null, especially in time-wise settings, and meaningful edits remain challenging. Our findings highlight the significant gap between the perceived and actual effectiveness of CodeLMs for real-world code update recommendation and emphasize the need for more research on improving their practicality, robustness, and generalizability.

cs.SE↗

arXiv2019

Unified model of sediment transport threshold and rate across weak and intense subaqueous bedload, windblown sand, and windblown snow

Nonsuspended sediment transport (NST) refers to the sediment transport regime in which the flow turbulence is unable to support the weight of transported grains. It occurs in fluvial environments (i.e., driven by a stream of liquid) and in aeolian environments (i.e., wind-blown) and plays a key role in shaping sedimentary landscapes of planetary bodies. NST is a highly fluctuating physical process because of turbulence, surface inhomogeneities, and variations of grain size and shape and packing geometry. Furthermore, the energy of transported grains varies strongly due to variations of their flow exposure duration since their entrainment from the bed. In spite of such variability, we here propose a deterministic model that represents the entire grain motion, including grains that roll and/or slide along the bed, by a periodic saltation motion with rebound laws that describe an average rebound of a grain after colliding with the bed. The model simultaneously captures laboratory and field measurements and discrete element method (DEM)-based numerical simulations of the threshold and rate of equilibrium NST within a factor of about 2, unifying weak and intense transport conditions in oil, water, and air (oil only for threshold). The model parameters have not been adjusted to these measurements but determined from independent data sets. Recent DEM-based numerical simulations (Comola, Gaume, et al., 2019, https://doi.org/10.1029/2019GL082195) suggest that equilibrium aeolian NST on Earth is insensitive to the strength of cohesive bonds between bed grains. Consistently, the model captures cohesive windblown sand and windblown snow conditions despite not explicitly accounting for cohesion.

physics.geo-ph↗