Speech Title: Machine Learning and Granular Computing as a Framework of Design of Models of Quantitative Software Engineering
It is evident that
with the omnipresence of software systems and their
diverse applications quite often concerning critical
environments, efficient models of software
artifacts, software attributes (for instance,
software complexity and maintainability) and
software processes (e.g., cost estimation) become of
paramount importance with far-reaching practical
impact. The nature of software data including their
limited number, sparsity, distributed nature and
associated with privacy and security constitutes a
remarkable challenge when engaging effectively
machine learning design framework.
We advocate that in light of the arguments raised above, it is highly desirable that suitable models of software products and processes: (i) are transparent and easily comprehended by users and stakeholders, (ii) exhibit logic fabric that contributes to the elevated level of interpretability, (iii) realize the design based on an orchestrated usage of data and available domain knowledge helpful to customize the key modeling agenda, and (iv) come with well-articulated credibility measures that help efficiently assess the relevance of the models, say associated quality of prediction outcomes.
It is demonstrated that the above-stated requirements are addressed by admitting a suitable level of abstraction of the model and the ensuing results. The detailed discussion is carried by concentrating on rule-based architectures. In this context, several representative conceptual and algorithmic developments in machine learning and Granular Computing are covered: federated learning, granular gradient boosting, and transfer learning (knowledge).
Biography: Witold Pedrycz (Life Fellow, IEEE) received the M.Sc. degree in computer science, the Ph.D. degree in computer engineering, and the D.Sci. degree in systems science from the Silesian University of Technology, Gliwice, Poland, in 1977, 1980, and 1984, respectively.,He is a Professor and the Canada Research Chair of computational intelligence with the Department of Electrical and Computer Engineering, University of Alberta, Edmonton, AB, Canada. He is also with the System Research Institute, Polish Academy of Sciences, Warsaw, Poland. He has authored 15 research monographs covering various aspects of computational intelligence, data mining, and software engineering. His current research interests include computational intelligence, fuzzy modeling, granular computing, knowledge discovery and data mining, fuzzy control, pattern recognition, knowledge-based neural networks, relational computing, and software engineering.,Prof. Pedrycz was a recipient of the IEEE Canada Computer Engineering Medal, the Cajastur Prize for Soft Computing from the European Center for Soft Computing, the Killam Prize, and the Fuzzy Pioneer Award from the IEEE Computational Intelligence Society. He is intensively involved in editorial activities. He is an Editor-in-Chief of Information Sciences, WIREs Data Mining and Knowledge Discovery (Wiley), and the International Journal of Granular Computing (Springer). He currently serves as a member of a number of editorial boards of other international journals and is a Former Editor-in-Chief of IEEE Transactions on Systems, Man, and Cybernetics—Part A: Systems and Humans. He is a Foreign Member of the Polish Academy of Sciences. He is a Fellow of the Royal Society of Canada.(Based on document published on 25 December 2020).
Biography: Shugong Xu is an IEEE Fellow，a professor at Shanghai University. In his 20+ years career in research (over 15 years in industrial research labs), he had over 50 issued US/WO/CN patents and published more than 150 peer-reviewed research papers. He was awarded "National Innovation Leadership Talent" from China government in 2013, IEEE Fellow in 2015. He also won 2017 Award for Advances in Communication from IEEE Communication Society. His current research interests include V2X, wireless communication systems, and machine learning etc.
Title: Preventing Concurrency Bugs through Temporal Coordination Abstractions
Abstract: In sequential systems, programmers are responsible for totally ordering events occurring in a system. This results in overly constraining when events may occur. In contrast, concurrent systems involve nondeterministic interleaving of actions at autonomous actors. Without additional constraints on the order of events at participating actors, an interleaving may lead to incorrect operation or deadlocks. The talk will describe constructs for temporal coordination of actors such as synchronization constraints, activators, session types, synchronizers and protocol description. I will then discuss the utility and limitations of these methods, including barriers to adoption. The talk will conclude with a perspective on open problems and research directions.
Biography: Dr. Gul Agha is Professor Emeritus and Research Professor of Computer Science at the University of Illinois at UrbanaChampaign, and CEO of Embedor Technologies. Agha is a Fellow of the ACM, and Fellow of the IEEE. He was a recepient of the 2019 ACM SigSoft Impact Paper Award. Dr. Agha served as Editor-in-Chief of IEEE Concurrency: Parallel, Distributed and Mobile Computing (1994-98), and of ACM Computing Surveys (2000-07). Dr. Agha is best known for his formalization of the Actor model which has been realized in industrial programming languages and frameworks such as Erlang, Scala/Akka, and Orleans. Agha and his research group developed Concolic Testing for programs with memory and concurrency. Concolic testing has been incorporated in industrial software testing tools such as KLEE, Microsoft SAGE, and S2E. Dr. Agha developed methods for Statistical Model Checking (SMC). SMC has been applied to biological systems and cyberphysical systems. Dr. Agha research also led to Euclidean model checking, a method to reason about the evolution of probability distributions. Other research contributions include methods to harness computational learning for program verification; logical methods for automated decentralized, predictive runtime verification of programs; and distributed algorithms for wireless sensor networks (WSNs). Dr. Agha co-founded Embedor Technologies which is applying WSNs to continually monitor the structural health of bridges, buildings and railroad tracks. Embedor's technology was used to monitor the world largest Ferris wheel during construction.
Title: Integrating Adaptability into Interactive Applications
Abstract: Adaptive user interfaces
are an alternative to the traditional
one-size-fits-all user interfaces. AUIs have the
ability to adapt their structures, appearances, and
behavior to a variety of objectives, aiming to
provide highly usable applications for people with
different needs and in different contexts of use.
Successful design and development of adaptive user
interfaces are one of the major research directions
in the areas of human computer interaction and
Noticeably the effectiveness of AUIs depends on how accurately adaptation satisfies user needs. As more information on the context of use is available, the AUI of an application evolves and changes in the application are inevitable. It is a great challenge to develop a reusable architecture to accommodate future changes.
Users make a sequence of decisions when they navigate a user interface. Such decisions are interdependent. Knowledge about activities that the user performs at runtime is crucial for adaptation decision making. It not only serves as a basis for evaluating relevance of the available information (such as user status, usage patterns, and context of use), but also facilitates reasoning about user needs. However, implementation of the user activity tracking capability often relies on intimate knowledge of the target application, which makes it difficult to develop loosely coupled modules to achieve reusability.
We propose a two-step approach to achieving reusability of software support for AUIs. Briefly, we use aspect-oriented instrumentation to capture user interface events and then conduct model-based analysis on event traces to identify user tasks. As our experiment shows, this approach provides reusable support for adaptability at the task level.
Biography: Dr. Yonglei Tao is a professor in the School of Computing and Information Systems at Grand Valley State University, Michigan, USA. He received his Ph.D. in Computer Science from the University of Iowa. His research interests include tool support for usability evaluation, software engineering, and computer science education.