University of Twente Student Theses
Investigating Approximate Computing to design an energy efficient deep learning architecture for anomaly detection from ECG signals
Capitani, Dario (2024) Investigating Approximate Computing to design an energy efficient deep learning architecture for anomaly detection from ECG signals.
|
PDF
712kB |
| Abstract: | Cardiovascular diseases make up 32% of global mortality, most of which are detectable through monitoring of physiological signals, such as Electrocardiograms (ECG). Through the use of an Anomaly Detection algorithm, deployed on a device that is continuously monitoring, such diseases can be detected before onset. The requirements for such a device are to be power efficient and small. Existing research covers models that are too large for a wearable device, or reduce the input size. The later allows for smaller models and hardware but leads to a loss of information. Herein, an adapted Multi Layer Perceptron is used taking a whole heart beat as an input, whose architecture is then developed using a Field Programmable Gate Array. From here existing techniques under the paradigm of Approximate Computing are applied, to further reduce hardware and power consumption. This proof of principle paper demonstrates the feasibility of achieving a potential diagnosis while reducing the hardware and power resources. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science BSc (56964) |
| Awards: | Best Paper Sustainable Artificial Intelligence |
| Link to this item: | https://purl.utwente.nl/essays/101025 |
| 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