Advancements in Knowledge Elicitation for Computer-based Critical Systems
摘要截稿:
全文截稿: 2019-02-15
影响因子: 6.125
期刊难度:
CCF分类: C类
中科院JCR分区:
• 大类 : 计算机科学 - 2区
• 小类 : 计算机:理论方法 - 1区
Overview
The availability of a huge amount of data has enabled the massive application of machine learning and deep learning techniques across the domain of computer-based critical systems. A huge set of automatic learning frameworks are now available, and are able to tackle with different kinds of systems, enabling the diffusion of Big Data analysis, cloud computing systems and (Industrial) Internet of Things. As such applications become more and more widespread, data analysis techniques have shown their capability to identify operational patterns and to predict future behaviour for anticipating possible problems.
Knowledge outcoming from these approaches are still hard to manipulate with high-level reasoning mechanisms (formal reasoning, model checking, model-based approaches): this special issue aims at exploring the synergy of model-based and data-driven approaches in order to boost critical applications and systems analysis and monitoring.
The advantages of such integration are numerous:
- to scale up complexity of data analysis allowing to reduce dimensionality in real world problems;
- to boost human activities in the supervision of complex system operations;
- to improve the trustworthiness of the system models built manually;
- to enhance the accuracy of the results predicted with the analysis. Inaccurate results may lead to invalid engineering and business decisions that have serious consequences in the critical domain;
- to support the creation ofmodels@runtimethat is to align models with data logged by the system in operation;
- to enable automatic validation of models extracted by data mining.
This special issue focuses on novel approaches, solutions and techniques able to combine the power of both model manipulation and data analysis. We are particularly interested in contributions that focus on model-driven approaches able to combine models and data according to different paradigms and techniques (e.g., data annotation, model learning, feature engineering). Such combination may be oriented to assessment/supervision of functional properties (e.g., correctness, absence of mis-behaviours) or quality of service (e.g., performance, dependability and cyber resilience).
We are interested in both theoretical papers and in practical applications of existing techniques to real scale case studies in application domains where physical world meet cutting-edge computing systems: (Industrial) Internet of Things, Big Data analysis architectures, Distributed Computing Systems, Smart Manufacturing and Smart Industrial settings, Intelligent Transportation Systems, Smart Cities, Smart Power Grids, Active and Assisted Living, etc.
Contributions must contain original unpublished work not concurrently submitted or under review anywhere elsewhere.
Topics of the special issue
Methodologies integrating/comparing modelling and data oriented approaches
Explicit knowledge elicitation from machine learning approaches
Data alignment formodels@runtime
Attack and failure pattern recognition in critical system
Model-driven reverse engineering applied to data analysis
Process/event mining based methods and approaches
Feature Engineering in functional and non-functional properties analysis
Multi-formalism modelling approaches and multi-solution analysis processes
Model-based machine learning
Model-learning approaches and techniques
Machine learning based architecture for control systems analysis
Advanced architectures for (industrial) Big Data management
Approaches for the assessment/supervision of the system quality of service (i.e., performance, dependability and cyber resilience) that integrate data and/or models