Is Machine Learning Software just Software: A Maintainability View

Authors: Tommi Mikkonen, Jukka Nurminen, Mikko Raatikainen, Ilenia Fronza, Niko Mäkitalo, Tomi Männistö

Artificial intelligence (AI) and machine learning (ML) is becoming commonplace in numerous fields. As they are often embedded in the context of larger software systems, issues that are faced with software systems in general are also applicable to AI/ML. In this paper, we address ML systems and their characteristics in the light of software maintenance and its attributes, modularity, testability, reusability, analysability, and modifiability. To achieve this, we pinpoint similarities and differences between ML software and software as we traditionally understand it, and draw parallels as well as provide a programmer's view to ML at a general level, using the established software design principles as the starting point.

Presented by: Tommi Mikkonen
Company: Department of Computer Science

Talk language: English
Level: Advanced
Target group:

Partner der Konferenz 2022

ASQF e.V ATB - Austrian Testing Board coderskitchen dpunkt.verlag GmbH Fortiss GmbH GTB - German Testing Board Heise Medien GmbH & Co. KG iSQI GmbH IT Verlag für Informationstechnik GmbH SIGS DATACOM GmbH TU Wien, Institut für Information Systems Engineering, CDL-SQI WKO - Wirtschaftskammer