Skip to main content

Compressing Deep Networks Using Fisher Score of Feature Maps

Publication ,  Conference
Soltani, M; Wu, S; Li, Y; Ravier, R; Ding, J; Tarokh, V
Published in: Data Compression Conference Proceedings
March 1, 2021

In this paper, we propose a new structural technique for pruning deep neural networks with skip-connections. Our approach is based on measuring the importance of feature maps in predicting the output of the model using their Fisher scores. These scores subsequently used for removing the less informative layers from the graph of the network. Extensive experiments on the classification of CIFAR-10, CIFAR-100, and SVHN data sets demonstrate the efficacy of our compressing method both in the number of parameters and operations.

Duke Scholars

Published In

Data Compression Conference Proceedings

DOI

ISSN

1068-0314

ISBN

9780738112275

Publication Date

March 1, 2021

Volume

2021-March

Start / End Page

371
 

Citation

APA
Chicago
ICMJE
MLA
NLM
Soltani, M., Wu, S., Li, Y., Ravier, R., Ding, J., & Tarokh, V. (2021). Compressing Deep Networks Using Fisher Score of Feature Maps. In Data Compression Conference Proceedings (Vol. 2021-March, p. 371). https://doi.org/10.1109/DCC50243.2021.00083
Soltani, M., S. Wu, Y. Li, R. Ravier, J. Ding, and V. Tarokh. “Compressing Deep Networks Using Fisher Score of Feature Maps.” In Data Compression Conference Proceedings, 2021-March:371, 2021. https://doi.org/10.1109/DCC50243.2021.00083.
Soltani M, Wu S, Li Y, Ravier R, Ding J, Tarokh V. Compressing Deep Networks Using Fisher Score of Feature Maps. In: Data Compression Conference Proceedings. 2021. p. 371.
Soltani, M., et al. “Compressing Deep Networks Using Fisher Score of Feature Maps.” Data Compression Conference Proceedings, vol. 2021-March, 2021, p. 371. Scopus, doi:10.1109/DCC50243.2021.00083.
Soltani M, Wu S, Li Y, Ravier R, Ding J, Tarokh V. Compressing Deep Networks Using Fisher Score of Feature Maps. Data Compression Conference Proceedings. 2021. p. 371.

Published In

Data Compression Conference Proceedings

DOI

ISSN

1068-0314

ISBN

9780738112275

Publication Date

March 1, 2021

Volume

2021-March

Start / End Page

371