Deploying TinyML models on low-cost IoT hardware is very challenging, due to limited device memory capacity. Neural processing unit (NPU) hardware address the memory challenge by using model compression to exploit weight quantization and sparsity to fit more parameters in the same footprint. However, designing compressible neural networks (NNs) is challenging, as it expands the design space across which we must make balanced trade-offs. This paper demonstrates Unified DNAS for Compressible (UDC) NNs, which explores a large search space to generate state-of-the-art compressible NNs for NPU. ImageNet results show UDC networks are up to 3.35× smaller (iso-accuracy) or 6.25% more accurate (iso-model size) than previous work.
@article{arxiv.2201.05842,
title = {UDC: Unified DNAS for Compressible TinyML Models},
author = {Igor Fedorov and Ramon Matas and Hokchhay Tann and Chuteng Zhou and Matthew Mattina and Paul Whatmough},
journal= {arXiv preprint arXiv:2201.05842},
year = {2023}
}