Academic Partner: School of Engineering at Jönköping University
Project duriation: 2022-2024
Explainable AI for product and service development
Artificial intelligence (AI) is becoming an increasing part of our lives and is predicted to play a major role in the manufacturing industry. XAI-Pro focuses on how to deliver new knowledge learned from existing data in an understandable and explainable way when making a recommendation or prediction during the development of products and services processes.
We carry out research in collaboration with Husqvarna AB and Saab AB Training & Simulation within two case studies, predictive maintenance and operational guidance.
XAI-Pro is one of several projects within the research profile AFAIR that applies AI in industrial organizations. XAI-Pro is part of the theme “Data driven development of products and services” (ARA1).
Motivation and purpose
The development of products and their daily use generate huge amounts of data. Certainly, such data includes valuable information about errors, preferences, omissions, as well as unintended usage which could be leveraged to improve these products and services. To some extent, machine learning (ML) may be able to identify interesting patterns and relationships that can provide decision support for product and service developers.
However, many of these ML systems often operate autonomously as “black boxes” that are not designed for transparent interaction with end users. Users thus have difficulties understanding the behaviour which can result in mistrust and misuse of these systems.
To mitigate some of the challenges connected to black-box systems, transparency, interpretability and explainable methods can be used. Explainable AI (XAI) has recently seen an increase in interest from the AI research community
We build explainable and interpretable ML models and methods for enhanced product and service development. We employ real-world use cases from our industrial partners, Husqvarna and Saab, and we build proof-of-concept prototypes to demonstrate the results.
For more information, contact:
Are you interested in a future collaboration?