Embedded Machine Learning
The embedded machine learning group works on the theory, methodology and optimization of deep learning methods as well as their applications on resource-constrained devices in real-world deployments.
In particular, we are interested in the following subjects:
- model compression methods for efficient on-device inference
- new methodologies for efficient on-device learning
- on-device multi-task learning
- applications of Data-driven methods in predicting and understanding air pollution
- local pre-processing of data at the edge for environmental monitoring