TY - GEN
T1 - An Enhanced Hybrid MobileNet
AU - Chen, Hong Yen
AU - Su, Chung Yen
N1 - Publisher Copyright:
© 2018 IEEE.
PY - 2018/10/31
Y1 - 2018/10/31
N2 - Complicated and deep neural network models can achieve high accuracy for image recognition. However, they require a huge amount of computations and model parameters, which are not suitable for mobile and embedded devices. Therefore, MobileNet was proposed, which can reduce the number of parameters and computational cost dramatically. The main idea of MobileNet is to use a depthwise separable convolution. Two hyper-parameters, a width multiplier and a resolution multiplier are used to the trade-off between the accuracy and the latency. In this paper, we propose a new architecture to improve the MobileNet. Instead of using the resolution multiplier, we use a depth multiplier and combine with either Fractional Max Pooling or the max pooling. Experimental results on CIFAR database show that the proposed architecture can reduce the amount of computational cost and increase the accuracy simultaneously.
AB - Complicated and deep neural network models can achieve high accuracy for image recognition. However, they require a huge amount of computations and model parameters, which are not suitable for mobile and embedded devices. Therefore, MobileNet was proposed, which can reduce the number of parameters and computational cost dramatically. The main idea of MobileNet is to use a depthwise separable convolution. Two hyper-parameters, a width multiplier and a resolution multiplier are used to the trade-off between the accuracy and the latency. In this paper, we propose a new architecture to improve the MobileNet. Instead of using the resolution multiplier, we use a depth multiplier and combine with either Fractional Max Pooling or the max pooling. Experimental results on CIFAR database show that the proposed architecture can reduce the amount of computational cost and increase the accuracy simultaneously.
KW - MobileNet
KW - deep learning
KW - image classifier
KW - image recognition
KW - neural networks
UR - http://www.scopus.com/inward/record.url?scp=85057433621&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85057433621&partnerID=8YFLogxK
U2 - 10.1109/ICAwST.2018.8517177
DO - 10.1109/ICAwST.2018.8517177
M3 - Conference contribution
AN - SCOPUS:85057433621
T3 - 2018 9th International Conference on Awareness Science and Technology, iCAST 2018
SP - 308
EP - 312
BT - 2018 9th International Conference on Awareness Science and Technology, iCAST 2018
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 9th International Conference on Awareness Science and Technology, iCAST 2018
Y2 - 19 September 2018 through 21 September 2018
ER -