What is it about?
To automatically monitor a production line with multiple sensors, we use deep learning to predict the expected range of upcoming sensor readings just before they're actually recorded. Then, as soon as a new reading comes in, we compare it to the predicted range. This allows us to instantly spot any unusual readings, helping detect issues in real time with very little delay.
Featured Image
Photo by Jakub Żerdzicki on Unsplash
Perspectives
I hope to highlight the importance of making deep learning techniques practical by focusing on three key factors: the amount of data needed for effective training, the hardware requirements for running the model, and how useful the model’s outputs are for real-world decisions, characteristics which may be lost when evaluating performance metrics but which are crucial in allowing these techniques to eventually leave the realm of research
Alessio Mascolini
Politecnico di Torino
Read the Original
This page is a summary of: VARADE: a Variational-based AutoRegressive model for Anomaly Detection on the Edge, June 2024, ACM (Association for Computing Machinery),
DOI: 10.1145/3649329.3655691.
You can read the full text:
Contributors
The following have contributed to this page
