GlucoNet applies feature decomposition and knowledge distillation to a transformer model to forecast blood glucose levels from irregular multimodal data, reporting accuracy gains and model compression on data from 12 T1D participants.
Efficient acceleration of deep learning inference on resource-constrained edge devices: A review,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Hybrid Attention Model Using Feature Decomposition and Knowledge Distillation for Glucose Forecasting
GlucoNet applies feature decomposition and knowledge distillation to a transformer model to forecast blood glucose levels from irregular multimodal data, reporting accuracy gains and model compression on data from 12 T1D participants.