University of Twente Student Theses
Supervised contrastive learning to overcome inconsistencies in exhaled breath data
Lucas, R.H. (2022) Supervised contrastive learning to overcome inconsistencies in exhaled breath data.
|
PDF
6MB |
| Abstract: | Disease prediction can be performed based on breath analysis through the recognition of Volatile Organic Compound patterns. Such data is acquired through clinical studies where patients breathe into an electronic nose. This gives temporal data for every patient. Making disease predictions based on this data is challenging as the data is sparse, temporally complex, possibly non-linear, and contains inconsistencies. Supervised Contrastive learning tackles the device- and/or location-based inconsistencies that reside within the data by learning general features of the data through the recognition of similarities and differences between data points. This research shows that Supervised Contrastive learning learns more effective data representations to be used for disease classification on never before seen devices and thereby overcoming the inconsistencies. In doing so a more generalized classifier has been realized that can be used across multiple devices and locations. |
| Item Type: | Essay (Master) |
| Clients: | The eNose Company, Zutphen, The Netherlands |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science MSc (60300) |
| Link to this item: | https://purl.utwente.nl/essays/92848 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page