Data Mining and Machine Learning for System Model Discovery and Applications

用于系统模型发现和应用的数据挖掘和机器学习

• 批准号: RGPIN-2019-05163
• 负责人: Narayan, Apurva
• 金额: $ 1.68万
• 依托单位: University of British Columbia
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2021
• 资助国家: 加拿大
• 起止时间: 2021-01-01 至 2022-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=738096
• 关键词: Data; Mining; Machine; Learning; System

Current day software infrastructures, cyber-physical systems, and Internet of Things (IoT) have become extremely complex for engineers to understand and debug. The risks are higher if the system is safety-critical. The increase in complexity of software systems presents a challenging problem of providing assurances of their reliable operation. Moreover, the lack of good quality behavioral description (both at macro and micro levels of abstraction) of these systems is one of the major problems. Due to their inherent complexity, and rapid evolution of software systems to meet the demands of clients, software applications and libraries are often released without documented specifications. Furthermore, lack of specifications negatively impacts the maintainability and reliability of systems. My research program, `Data Mining and Machine Learning for System Model Discovery and Applications', focuses on developing novel, faster, intelligent algorithms, and frameworks, and on seeing their broader impact through practical applications. The primary goal of my research program is to develop algorithms, models, and tools that infer the model/behavior of complex software systems considering them as a 'black box'. I propose to concentrate on model inference of complex software systems (both ubiquitous and large scale infrastructures). Specifically, I aim to infer system behavior, visualize it, and use it to make complex systems understandable, robust and reliable. Models developed as a part of this research program will not only infer complex system behavior model, but also incorporate the results into an experimentation framework for system safety and security. Consequently, users will benefit from the latest results at runtime enabling safe, secure, and resilient cyber-physical systems. The outcome of this work will be a set of open source software tools that will help engineers to build and analyze systems quicker and with fewer mistakes. This work will be relevant to both industry and academics. It will solve everyday problems faced by software engineers as well as extend our understanding of how to reason the complex behavior of software infrastructures. Further, this research will provide important and invaluable training to both graduate and undergraduate students. It will produce highly-qualified personnel with `data mining and machine learning skill-set' that is valuable and actively sought after by all the major software, hardware, and consulting companies in Canada. Lastly, this research program will develop strong partnerships between my group with industries and academia.

当今的软件基础设施，网络物理系统和物联网（IoT）对于工程师来说已经变得非常复杂，难以理解和调试。如果系统是安全关键型的，则风险更高。软件系统复杂性的增加提出了一个具有挑战性的问题，即如何保证其可靠运行。此外，这些系统缺乏高质量的行为描述（在宏观和微观抽象层面）是主要问题之一。由于其固有的复杂性，以及软件系统的快速发展，以满足客户的需求，软件应用程序和库往往没有文档化的规范发布。此外，缺乏规范对系统的可维护性和可靠性产生负面影响。我的研究项目，“数据挖掘和机器学习系统模型发现和应用”，专注于开发新颖，更快，智能算法和框架，并通过实际应用看到其更广泛的影响。我的研究计划的主要目标是开发算法，模型和工具，推断复杂软件系统的模型/行为，将它们视为“黑匣子”。我建议专注于复杂软件系统（包括无处不在的和大规模的基础设施）的模型推理。具体来说，我的目标是推断系统行为，将其可视化，并使用它来使复杂的系统变得可理解，健壮和可靠。作为该研究计划的一部分开发的模型不仅可以推断复杂的系统行为模型，还可以将结果纳入系统安全性和安全性的实验框架。因此，用户将受益于运行时的最新结果，从而实现安全、可靠和有弹性的网络物理系统。这项工作的成果将是一套开源软件工具，帮助工程师更快地构建和分析系统，减少错误。这项工作将与工业界和学术界有关。它将解决软件工程师面临的日常问题，并扩展我们对如何推理软件基础设施复杂行为的理解。此外，这项研究将提供重要的和宝贵的培训，研究生和本科生。它将培养出具有“数据挖掘和机器学习技能”的高素质人才，这些人才是加拿大所有主要软件，硬件和咨询公司所积极追求的。最后，这个研究项目将在我的团队与工业界和学术界之间建立牢固的伙伴关系。