paperbot · PL 论文追踪

全部论文 待读 收藏

P2 MARQ: Engineering Mission-Critical AI-Based Software with Automated Result Quality Adaptation

MARQ:面向任务关键型 AI 软件的自动化结果质量适配工程
ICSE 2025 IEEE/ACM 47th International Conference on Software Engineering (ICSE)2025匹配度 2/10 · 该论文关注 AI 服务链的运行时 QoR 与资源约束优化,并提出 DSL 建模,与用户核心方向(静态分析、Agentic 程序分析、浮点/贝叶斯程序分析与 LLM 交叉)几乎没有直接关联。虽然标签涉及程序语言与机器学习,但属于运行时自适应与软件工程,而非静态分析或形式化验证。引用 2 inbox
Uwe Gropengießer, Elias Dietz, Florian Brandherm, Achref Doula, Osama Abboud, Xun Xiao, Max Mühlhäuser
crossref: 2026-09-12

AI 速览

一句话:提出 MARQ 框架,让开发者声明不同 QoR 的 AI 服务链,并在运行时自动选择以满足实时约束。

任务关键型 AI 软件具有统计性,结果质量(QoR)可调,但必须满足执行截止期等严格约束;移动与边缘场景中网络和资源波动使权衡更复杂。MARQ 允许软件工程师声明多条具有不同 QoR 与资源需求的候选 AI 服务链,每条链由微服务封装的 AI 服务组成。框架在运行时根据当前资源状况自动进行动态优化与链选择。评估表明,MARQ 能在实时条件下有效选择服务链并维持应用约束,服务链选择加速达 100 倍,QoR 较现有方法平均提升 10%。

问题
如何在资源波动(如移动/边缘网络)下,让任务关键型 AI 软件在满足执行截止期等严格约束的同时,灵活权衡并优化结果质量 QoR。
方法
MARQ 让工程师声明多条 QoR 与资源需求不同的 AI 服务链(微服务封装的 AI 服务序列),并在运行时自动进行动态优化与选择。
评估
在实时场景中评估 MARQ 的服务链选择效果;结果显示选择速度提升 100 倍,QoR 平均提升 10%,并能维持任务关键型应用的约束。
原文摘要(Abstract)

AI-based mission-critical software exposes a blessing and a curse: its inherent statistical nature allows for flexibility in result quality, yet the mission-critical importance demands adherence to stringent constraints such as execution deadlines. This creates a space for trade-offs between the Quality of Result (QoR)-a metric that quantifies the quality of a computational outcome-and other application attributes like execution time and energy, particularly in real-time scenarios. Fluctuating resource constraints, such as data transfer to a remote server over unstable network connections, are prevalent in mobile and edge computing environments-encompassing use cases like Vehicle-to-Everything, drone swarms, or social-VR scenarios. We introduce a novel approach that enables software engineers to easily specify alternative AI service chains-sequences of AI services encapsulated in microservices aiming to achieve a predefined goal-with varying QoR and resource requirements. Our methodology facilitates dynamic optimization at runtime, which is automatically driven by the MARQ framework. Our evaluations show that MARQ can be used effectively for the dynamic selection of AI service chains in real-time while maintaining the required application constraints of mission-critical AI software. Notably, our approach achieves a 100x acceleration in service chain selection and an average 10% improvement in QoR compared to existing methods.

链接与引用

DOI 原文 ·

BibTeX
@inproceedings{paperbot8916,
  title = {MARQ: Engineering Mission-Critical AI-Based Software with Automated Result Quality Adaptation},
  author = {Uwe Gropengießer and Elias Dietz and Florian Brandherm and Achref Doula and Osama Abboud and Xun Xiao and Max Mühlhäuser},
  booktitle = {ICSE},
  year = {2025},
  doi = {10.1109/icse55347.2025.00182}
}