Robust Self Supervised Model Development Stratio Automotive - ELISE Project Open Call 1
from Finnish Center for Artificial Intelligence FCAI Hosted by DASER
Stratio develops models to predict vehicle faults before they occur, so that the driver can intervene before a breakdown occurs. The main objective of this project was to understand if it is possible to shorten the development time of these models by using a robust self-supervised approach. We hypothesized that this would be achieved by automating the information retrieval task from raw vehicle data, that is typically done by a domain expert. The results show some significant success, although still with performance below the manual alternative. Further investments will be done to optimize performance. A model developed with this approach was deployed, providing an additional source of information for our analysts.
Embed this episode
No transcript yet.