The seminar series includes 4 talks of 1h and 15min each, followed by a Discussion, during the same day 03.12.2021. The exact time of each talk, as well as an introductory abstract, are detailed below.
First Talk
9:00 -10:15 (UTC +1)
By Prof. Anupam CHATTOPADHYAY (Nanyang Technological University, Singapore)
Abstract: In recent years, rapid advances in machine intelligence have tremendously boosted the capabilities of Internet-of-Things (IoT) platforms. This opened up doors for innovations in diverse non-digital industries, which can reap the benefits of distributed intelligence and faster communication to address problems like robust system design, predictive maintenance, remote asset monitoring and in general support the components of industry 4.0. However, the devils are in the details. In this talk, we will go through some practical issues faced by researchers while setting up a large-scale AI-powered IoT platform. We will discuss case studies from several industries undertaken through an IoT/Data analytics spin-off.
Second Talk
10:45 -12:00 (UTC+1)
By Prof. Jeronimo CASTRILLON (TU Dresden, Germany)
Abstract: Embedded and cyber-physical systems (CPS) are heterogeneous interconnected computing systems with an ever increasing complexity. As CPSs become more widespread, the developer community widens, exposing the complexity to mainstream programmers. In this talk, we will discuss domain-specific languages (DSLs) as a promising avenue to handle complexity without compromising on efficiency. The talk will motivate the need for DSLs and present a couple of prominent examples to then focus on recent results targeting emerging computing systems. Finally, we will briefly touch upon the MLIR project, a compiler framework for the development and reuse of abstractions and optimizations across DSLs.
Third Talk
14:00 -15:15 (UTC+1)
By Prof. Sanjoy BARUAH (Washington University in ST. Louis, USA)
Abstract: Many safety-critical Cyber-Physical Systems are required to have their safety properties validated prior to deployment. However, a good deal of uncertainty often characterizes the interactions of complex CPSs with the physical world within which they operate: validating correctness is typically done based on worst-case assumptions regarding such interactions and tends to result in resource-inefficient system implementations. We will investigate approaches for characterizing these interactions in a manner that enables more resource-efficient implementation while remaining compliant with safety requirements.
Fourth Talk
15:30 - 16:45 (UTC+1)
By Prof. Edward A. LEE (University of California at Berkeley, USA)
Abstract: In applications such as industrial automation and autonomous vehicles, computational systems interact tightly with physical processes. The design of such systems involves both scientific methods and engineering methods. In this talk, I examine from a high level the distinctions between these methods, particularly in their use of models. In science, the value of a model lies in how well it emulates a physical system. In engineering, the value of a physical system lies in how well it emulates a model. Since complex systems invariably involve a multiplicity of models, the relation between models becomes as important as the relation between models and physical systems. Relations between models are the domain of mathematics and underly both synthesis and analysis techniques in engineering. From this world view, I derive six design principles and describe how my current research projects leverage these principles.
Discussion
16:45-17:30 (UTC+1)