Motivation

Over the last decade, artificial intelligence (AI) and specifically machine- and deep-learning
(ML/DL) solutions have grown exponentially in prominence. Because of the big data era, and with
companies collecting customer and product data from an increasing number of connected devices, more
data is available than ever before and can be used for analytics, e.g. A/B testing, as well as training
ML/DL solutions. In parallel, the progress in high-performance parallel hardware, such as GPUs, TPUs
and FPGAs, allows for training solutions of scales unfathomable even a decade ago. These two
concurrent technology developments are at the heart of the rapid adoption of data-driven practices and
ML/DL solutions in industry.

The hype around AI has resulted in virtually every company having some form of AI initiative, or host of AI
initiatives, ongoing and the number of experiments and prototypes in industry is phenomenal. However,
research shows that the transition from prototype to industry-strength, production-quality deployment of
ML/DL models proves to be challenging for many companies. The engineering challenges, and the
related data management challenges, prove to be significant even if many data scientists and companies
fail to recognize these.

One of the important application domains for AI is software engineering. AI can be used for, among
others, software analytics, quality assurance and testing, software processes, software visualization,
human-computer interaction, adaptive systems and data management. AI thus enables us to take the
next step in excelling in software development and operations by delivering smarter software monitoring,
analytics, development and management techniques, methods and tools. Specifically, DevOps, due to
the periodic nature and significant amounts of data generated, is highly suitable for the application of
ML/DL models.
This track is concerned with novel research results in the area of engineering to facilitate production-
quality AI solutions in all relevant domains, including software engineering.

Topics

Topics of interest include, but are not restricted to:

● Solutions to assess and guarantee data quality for M, including data pipelines
● Design methods and approaches for developing ML/DL models
● Distributed ML/DL models in embedded systems, including federated learning and distributed AI
● Methods, tools, applications and lessons learned of AI for software engineering
● Automated labeling of data for ML
● Adoption of DevOps, DataOps, and/or MLOps practices in large-scale software engineering
● AI and analytics for DevOps
● Engineering aspects of training, transfer learning and reinforcement learning
● Engineering effective ML/DL deployments
● Automated experimentation and autonomously improving systems
● Feature experimentation and data driven development practices (e.g., A/B testing)
● Reinforcement learning and multi-armed bandits
● Explainable and compliant AI in the context of DevOps
● Predictive methods and estimation in software development, operations, and evolution
● Software visualization and visual analytics
● Empirical studies and experience reports about successful or unsuccessful applications of the aforementioned topics

In particular, we encourage submissions demonstrating the benefits and/or challenges with regards to the
development, deployment and evolution of the technologies mentioned above – as well as the adoption
and application of the practices, tools and techniques related to these. We welcome submissions
providing empirical case study data to illustrate how companies approach this shift in development
paradigms.

Track/Session Organizers

Helena Holmström Olsson, helena.holmstrom.olsson@mau.se, Malmö University, Sweden
Jan Bosch, jan.bosch@chalmers.se, Chalmers University of Technology, Sweden

Program Committee

  • Ankit Agrawal, University of Notre Dame
  • Matteo Camilli, Politecnico di Milano
  • Daniela S. Cruzes, NTNU
  • Aleksander Fabijan, Microsoft
  • Ilias Gerostathopoulos, Vrije Universiteit Amsterdam
  • Karthik Vaidhyanathan, IIIT Hyderabad
  • Hans-Martin Heyn, University of Gothenburg
  • Dietmar Winkler, Vienna University of Technology
  • Henry Muccini, University of L’Aquila
  • Sami Hyrynsalmi, LUT University
  • Maya Daneva, University of Twente
  • Tommi Mikkonen, University of Helsinki
  • Philipp Haindl, St. Pölten University of Applied Sciences
  • Michel Chaudron, Eindhoven University of Technology
  • Michael Felderer, German Aerospace Center (DLR)
  • Justus Bogner, University of Stuttgart
  • Jens Heidrich, Fraunhofer
  • Christoph Elsner, Siemens AG
  • Rudolf Ramler, Software Competence Center Hagenberg
  • Sandro Morasca, Università degli Studi dell’Insubria
  • Xiaofeng Wang, Free University of Bozen-Bolzano
  • Matthias Tichy, Ulm University
  • Andriy Miranskyy, Ryerson University
  • Stefan Wagner, University of Stuttgart
  • Ezequiel Scott, University of Tartu
  • Zoltan Mann, University of Amsterdam
  • Matthias Galster, University of Canterbury
  • Michael Klaes, Fraunhofer
  • Steffen Frey, University of Groningen
  • Nazim Madhavji, University of Western Ontario
  • Luis Cruz, Delft University of Technology
  • Ipek Ozkaya, Carnegie Mellon University Software Engineering Institute
  • Andreas Metzger, University of Duisburg-Essen
  • Deepanjan Kundu, Meta
  • Eduardo Guerra, Free University of Bozen-Bolzano