MicroT: Low-Energy and Adaptive Models for MCUs

arXiv:2403.08040v1 Announce Type: new
Abstract: We propose MicroT, a low-energy, multi-task adaptive model framework for resource-constrained MCUs. We divide the original model into a feature extractor and a classifier. The feature extractor is obtained through self-supervised knowledge distillation and further optimized into part and full models through model splitting and joint training. These models are then deployed on MCUs, with classifiers added and trained on local tasks, ultimately performing stage-decision for joint inference. In this process, the part model initially …

abstract arxiv classifier cs.ar cs.lg distillation energy feature framework knowledge low low-energy mcus part through training type