Data e hora: terça, 24 de outubro de 2023, às 17h.
Local: B307 – CAD3
Palestrante: João R. Campos, University of Coimbra
Título: Intelligent Dependability
Abstract
Critical and sensitive tasks are executed through complex software systems on a daily basis. Issues or failures in such systems can easily lead to considerable financial losses or even loss of lives. Notwithstanding, software size and complexity have been growing considerably, reaching proportions that render traditional validation techniques impractical. It has become virtually impossible to detect all software faults before deployment which may eventually lead to failures at runtime or leave the system open for exploitation. Various techniques have been developed with the purpose of avoiding and handling such situations, but this still remains an open issue. Machine Learning (ML) algorithms have shown their ability to adapt and extract knowledge in a variety of complex problems. They are able to find intricate patterns in the data and learn from them, often without assumptions about the underlying model. Thus, it presents a high potential to create novel techniques to improve the dependability of modern systems. Still, each problem has different characteristics that require different techniques, and their choice and optimization is in itself complex.
In this talk I will focus on the research topics that I have been working on, as well as future research directions. Currently, my work focuses on studying, adapting, and applying the state of the art of Artificial Intelligence (AI) and ML within the field of Dependability, promoting the field of Intelligent Dependability. Despite the ubiquity of ML nowadays, and the “plug-n-play” nature of some ML packages, properly devising AI-based solutions to achieve more dependable systems requires specific considerations, tools, and processes, as each problem is intrinsically different. Thus, the current main goal is to explore state-of-the-art techniques to develop novel methods, or support existing techniques, to improve the dependability of modern complex systems. Assuring the performance of those systems is also a challenging open issue as although ML has achieved remarkable results, it has also been shown that specific inputs cause unexpected results. This research tries to go beyond merely using AI/ML algorithms, it focuses on also devising novel processes and approaches that take into consideration the specific characteristics of the target domain. Notwithstanding, this requires expert knowledge and training in interdisciplinary fields.
Bio
João R. Campos is an Assistant Professor at the Department of Informatics Engineering at the University of Coimbra. His main research interests are on the use of Machine Learning (ML) to improve the dependability of modern complex systems, more precisely on the intersection between ML and Dependability, promoting the field of Intelligent Dependability. He works within the Software and Systems Engineering (SSE) group, focusing on how to leverage ML to solve relevant complex problems, from fault injection to failure prediction and the dependability of autonomous systems. He also collaborates in various projects using AI in different domains, from sports and health to astrophysics. He has participated in various Horizong and national projects, mainly on the intersection of ML and Dependability.
