Compact and Fast Machine Learning Accelerator for IoT Devices by Hantao Huang & Hao Yu

Author:Hantao Huang & Hao Yu
Language: eng
Format: epub
ISBN: 9789811333231
Publisher: Springer Singapore

Activation matrix of d layers

Labels and neural network output

SVD decomposition matrices

Input features

Rank of tensor cores

Index vectors referring to tensor element

Mode size of tensor

Predicted probability on each class

Maximum mode size of

Number of training samples

N

Number of input features

M

Number of classes

Tucker decomposition [9, 52] is a high-order singular value decomposition (SVD) and very stable. However, it is mainly designed for the 3-dimensional matrix and the number of parameters is relatively large, . Therefore, this decomposition method is not considered for model compression.

A d-dimensional tensor is decomposed into the tensor-train data format if tensor core is defined as and each element is defined [52] as

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.