jan 15, 2015 - Compression of deep convolutional neural networks for fast and low power mobile applications. (ICLR 2015)

Description:

Three-component Decomposition

以较小的精度损失为代价，显著降低了模型尺寸、运行时间和能耗。

w*h*c*n -> (spatial size 1*1) w*h 1*1

The proposed scheme consists of three steps: (1) rank selection with variational Bayesian ma- trix factorization, (2) Tucker decomposition on kernel tensor, and (3) fine-tuning to recover accumulated loss of accuracy, and each step can be easily implemented using publicly available tools.

结果：
Significant reductions in model size, runtime, and energy consumption are obtained, at the cost of small loss in ac- curacy.

Added to timeline:

Date: