Industrial-grade Verification & Validation of Evolving Systems - Canadian cluster

进化系统(ESS)形成了一类新的系统，由于开发中的快速迭代周期和/或它们的自适应和学习能力，可以迅速改变它们的行为；它们可以包括支持机器学习(ML)的组件。到2025年，加拿大人工智能相关产品的市场规模预计将达到1270亿美元。在加拿大和全世界，基于ML的(关键)应用程序的占用空间正在急剧增加。关键领域包括但不限于航空航天、网络安全、汽车和交通、银行和金融以及商业和数据分析。ES不遵循为传统系统开发的建模、设计和实施标准：ES的行为不像大多数传统系统那样透明；ES的开发是灵活的，通常由数据或用户不断变化的预期驱动。ESS的正式规范可能会缺失，而用数据或非结构化文本表达的需求往往是可用的。行业和我们的安全需要工业级测试和验证(T&V)方法，以全面和彻底地保证ESS的质量。加拿大研究机构CRIM和三个加拿大工业组织联合了来自四个欧洲国家的20多个组织，以满足IVVES(演进系统的工业级验证和确认)项目的工业需求。IVVES已获得尤里卡集群ITEA的批准。这项在IVVES-CC内进行的为期3年的研究将提出系统的方法，以实现ESS稳健而全面的工业级T&V。它开发了(1)致力于支持ML的应用的T&V方法，(2)致力于ES的人工智能和数据驱动的T&V方法，以涵盖目前最先进的基于规范的测试无法覆盖的领域，以及(3)使用数据分析为ES建立高效和高质量工程流程的智能工程方法。这些方法将根据工业使用案例进行评估；HQP和工业合作伙伴将接受培训，以掌握为更好地在该行业推出而开发的方法。该项目得到了CRIAQ的部分支持。

Evolving systems (ESs) form a new class of systems that can rapidly change their behavior, due to fast iteration cycles in development and/or to their ability to self-adapt and learn; they can include Machine Learning (ML)-enabled components. By 2025 the market for artificial intelligence related products in Canada is expected to reach $127 billion. The footprint of ML-based (critical) applications is dramatically increasing in Canada and worldwide. Critical domains include but are not limited to Aerospace, Cybersecurity, Automotive & Transportation, Banking & Finance and Business and data analytics. ESs do not follow the modelling, design and implementation standards developed for traditional systems: the behavior of ES is less transparent than for most of traditional systems; the development of ESs is agile and often driven by data or user-evolving expectations. Formal specifications for ESs can be missing and requirements expressed with data or unstructured text are often available. Industrial-grade testing and verification (T&V) approaches for a comprehensive and thorough quality assurance of ESs are needed by the industry and for our safety. CRIM, a Canadian research organisation, and three Canadian industrial organisations teamed up over 20 organisations from four European countries to fulfil the industrial needs within the context of the project IVVES (Industrial-graded Verification and Validation of Evolving Systems). IVVES has been approved by ITEA that is a cluster of EUREKA. The research conducted for 3 years within IVVES-CC will propose systematic approaches for robust and comprehensive, industrial-grade T&V of ESs. It develops (1) T&V approaches dedicated to ML-enabled applications, (2) AI and data-driven T&V approaches dedicated to ESs to cover areas that currently cannot be covered by state-of-the-art specification-based testing and (3) smart engineering approaches that use data analytics to establish efficient and high-quality engineering processes for ESs. The approaches will be evaluated on industrial use cases; HQP and industrial partners will be trained to master the developed approaches for a better roll-out in the industry. The project is partially supported by CRIAQ.