BriefGPT.xyz
Sep, 2017
领域自适应的深度网络压缩
Domain-adaptive deep network compression
HTML
PDF
Marc Masana, Joost van de Weijer, Luis Herranz, Andrew D. Bagdanov, Jose M Alvarez
TL;DR
本文研究深度神经网络在领域迁移后的压缩问题,提出一种基于低秩矩阵分解的压缩算法,结合目标域的激活统计信息,在不降低模型性能的情况下将模型参数量压缩至传统技术的5-20%。
Abstract
deep neural networks
trained on large datasets can be easily transferred to new domains with far fewer labeled examples by a process called
fine-tuning
. This has the advantage that representations learned in the
→