May 25, 2021 · Experimental results show that using generators discovered by the AutoRecon method always improve the performance of data-free compression.

Data-free compression raises a new challenge be- cause the original training dataset for a pre-trained model to be compressed is not available due to.

This work is the first to consider network engineering as an approach to design the reconstruction method, and results show that using generators discovered ...

May 25, 2021 · Specifically, we propose the AutoReCon method, which is a neural architecture search-based reconstruction method. In the proposed AutoReCon ...

@inproceedings{ijcai2021p478, title = {AutoReCon: Neural Architecture Search-based Reconstruction for Data-free Compression}, author = {Zhu, Baozhou and ...

The comparison between the current reconstruction method and the AutoReCon method for data-free compression. The goal of every subfigure is to update the ...

In this paper, we propose a strategy, named NASB, which adopts Neural Architecture Search (NAS) to find an optimal architecture for the binarization of CNNs.

AutoReCon: Neural Architecture Search-based Reconstruction for Data-free Compression. Baozhou Zhu, Peter Hofstee, Johan Peltenburg, Jinho Lee, Zaid Al-Ars

AutoReCon: Neural architecture search-based reconstruction for data-free compression. B Zhu, P Hofstee, J Peltenburg, J Lee, Z Alars. arXiv preprint arXiv: ...

Thus, a common approach is to compute a reconstructed training dataset before compression. The current reconstruction methods compute the reconstructed training ...