Complexity of Deep Convolutional Neural Networks in Mobile Computing
Neural networks employ massive interconnection of simple computing units called neurons to compute the problems that are highly nonlinear and could not be hard coded into a program. These neural networks are computation-intensive, and training them requires a lot of training data. Each training exam...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2020-01-01
|
| Series: | Complexity |
| Online Access: | http://dx.doi.org/10.1155/2020/3853780 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832567301038145536 |
|---|---|
| author | Saad Naeem Noreen Jamil Habib Ullah Khan Shah Nazir |
| author_facet | Saad Naeem Noreen Jamil Habib Ullah Khan Shah Nazir |
| author_sort | Saad Naeem |
| collection | DOAJ |
| description | Neural networks employ massive interconnection of simple computing units called neurons to compute the problems that are highly nonlinear and could not be hard coded into a program. These neural networks are computation-intensive, and training them requires a lot of training data. Each training example requires heavy computations. We look at different ways in which we can reduce the heavy computation requirement and possibly make them work on mobile devices. In this paper, we survey various techniques that can be matched and combined in order to improve the training time of neural networks. Additionally, we also review some extra recommendations to make the process work for mobile devices as well. We finally survey deep compression technique that tries to solve the problem by network pruning, quantization, and encoding the network weights. Deep compression reduces the time required for training the network by first pruning the irrelevant connections, i.e., the pruning stage, which is then followed by quantizing the network weights via choosing centroids for each layer. Finally, at the third stage, it employs Huffman encoding algorithm to deal with the storage issue of the remaining weights. |
| format | Article |
| id | doaj-art-3e76441623df43afb072453ede90ef27 |
| institution | Kabale University |
| issn | 1076-2787 1099-0526 |
| language | English |
| publishDate | 2020-01-01 |
| publisher | Wiley |
| record_format | Article |
| series | Complexity |
| spelling | doaj-art-3e76441623df43afb072453ede90ef272025-02-03T01:01:52ZengWileyComplexity1076-27871099-05262020-01-01202010.1155/2020/38537803853780Complexity of Deep Convolutional Neural Networks in Mobile ComputingSaad Naeem0Noreen Jamil1Habib Ullah Khan2Shah Nazir3Department of Computer Science, National University of Computer and Emerging Sciences, Islamabad, PakistanDepartment of Computer Science, National University of Computer and Emerging Sciences, Islamabad, PakistanDepartment of Accounting & Information Systems, College of Business & Economics, Qatar University, Doha, QatarDepartment of Computer Science, University of Swabi, Swabi, PakistanNeural networks employ massive interconnection of simple computing units called neurons to compute the problems that are highly nonlinear and could not be hard coded into a program. These neural networks are computation-intensive, and training them requires a lot of training data. Each training example requires heavy computations. We look at different ways in which we can reduce the heavy computation requirement and possibly make them work on mobile devices. In this paper, we survey various techniques that can be matched and combined in order to improve the training time of neural networks. Additionally, we also review some extra recommendations to make the process work for mobile devices as well. We finally survey deep compression technique that tries to solve the problem by network pruning, quantization, and encoding the network weights. Deep compression reduces the time required for training the network by first pruning the irrelevant connections, i.e., the pruning stage, which is then followed by quantizing the network weights via choosing centroids for each layer. Finally, at the third stage, it employs Huffman encoding algorithm to deal with the storage issue of the remaining weights.http://dx.doi.org/10.1155/2020/3853780 |
| spellingShingle | Saad Naeem Noreen Jamil Habib Ullah Khan Shah Nazir Complexity of Deep Convolutional Neural Networks in Mobile Computing Complexity |
| title | Complexity of Deep Convolutional Neural Networks in Mobile Computing |
| title_full | Complexity of Deep Convolutional Neural Networks in Mobile Computing |
| title_fullStr | Complexity of Deep Convolutional Neural Networks in Mobile Computing |
| title_full_unstemmed | Complexity of Deep Convolutional Neural Networks in Mobile Computing |
| title_short | Complexity of Deep Convolutional Neural Networks in Mobile Computing |
| title_sort | complexity of deep convolutional neural networks in mobile computing |
| url | http://dx.doi.org/10.1155/2020/3853780 |
| work_keys_str_mv | AT saadnaeem complexityofdeepconvolutionalneuralnetworksinmobilecomputing AT noreenjamil complexityofdeepconvolutionalneuralnetworksinmobilecomputing AT habibullahkhan complexityofdeepconvolutionalneuralnetworksinmobilecomputing AT shahnazir complexityofdeepconvolutionalneuralnetworksinmobilecomputing |