Name
Abstract | ||
|---|---|---|
In deep neural networks (DNNs), by removing unnecessary subnetworks can reduce neural network computing redundancy, or by extracting sub-networks with features can explain how the DNN works. In this paper, we propose a neural network structure search algorithm based on evolutionary algorithm (EA) for DNN pruning and features separation, especially considering the evolutionary efficiency issues and individual evaluation. We get verified on the CIFAR-10 (C10) and CIFAR-100 (C100) datasets. With the classification accuracy with a variation of +/- 1% as the precondition, our method significantly reduces the amount of calculations.
|
Year | DOI | Venue |
|---|---|---|
2020 | 10.1145/3377929.3389970 | GECCO '20: Genetic and Evolutionary Computation Conference
Cancún
Mexico
July, 2020 |
DocType | ISBN | Citations |
Conference | 978-1-4503-7127-8 | 0 |
PageRank | References | Authors |
0.34 | 3 | 6 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Zhaoyang Wu | 1 | 0 | 0.34 |
| Lin Lin | 2 | 1 | 1.70 |
| Guoliang Gong | 3 | 0 | 0.34 |
| Rui Xu | 4 | 0 | 0.34 |
| Mitsuo Gen | 5 | 1873 | 130.43 |
| Yong Zhou | 6 | 0 | 0.34 |