Convolutional Neural Networks (CNNs) have shown remarkable performance in image recognition tasks. In this letter, we propose a new CNN model called the EnsNet which is composed of one base CNN and multiple Fully Connected SubNetworks (FCSNs). In this model, the set of feature maps generated by the last convolutional layer in the base CNN is divided along channels into disjoint subsets, and these subsets are assigned to the FCSNs. Each of the FCSNs is trained independent of others so that it can predict the class label of each feature map in the subset assigned to it. The output of the overall model is determined by majority vote of the base CNN and the FCSNs. Experimental results using the MNIST, Fashion-MNIST and CIFAR-10 datasets show that the proposed approach further improves the performance of CNNs. In particular, an EnsNet achieves a state-of-the-art error rate of 0.16% on MNIST.
Daiki HIRATA
Industrial Technology Center of Okayama Prefecture
Norikazu TAKAHASHI
Okayama University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Daiki HIRATA, Norikazu TAKAHASHI, "Ensemble Learning in CNN Augmented with Fully Connected Subnetworks" in IEICE TRANSACTIONS on Information,
vol. E106-D, no. 7, pp. 1258-1261, July 2023, doi: 10.1587/transinf.2022EDL8098.
Abstract: Convolutional Neural Networks (CNNs) have shown remarkable performance in image recognition tasks. In this letter, we propose a new CNN model called the EnsNet which is composed of one base CNN and multiple Fully Connected SubNetworks (FCSNs). In this model, the set of feature maps generated by the last convolutional layer in the base CNN is divided along channels into disjoint subsets, and these subsets are assigned to the FCSNs. Each of the FCSNs is trained independent of others so that it can predict the class label of each feature map in the subset assigned to it. The output of the overall model is determined by majority vote of the base CNN and the FCSNs. Experimental results using the MNIST, Fashion-MNIST and CIFAR-10 datasets show that the proposed approach further improves the performance of CNNs. In particular, an EnsNet achieves a state-of-the-art error rate of 0.16% on MNIST.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2022EDL8098/_p
Copy
@ARTICLE{e106-d_7_1258,
author={Daiki HIRATA, Norikazu TAKAHASHI, },
journal={IEICE TRANSACTIONS on Information},
title={Ensemble Learning in CNN Augmented with Fully Connected Subnetworks},
year={2023},
volume={E106-D},
number={7},
pages={1258-1261},
abstract={Convolutional Neural Networks (CNNs) have shown remarkable performance in image recognition tasks. In this letter, we propose a new CNN model called the EnsNet which is composed of one base CNN and multiple Fully Connected SubNetworks (FCSNs). In this model, the set of feature maps generated by the last convolutional layer in the base CNN is divided along channels into disjoint subsets, and these subsets are assigned to the FCSNs. Each of the FCSNs is trained independent of others so that it can predict the class label of each feature map in the subset assigned to it. The output of the overall model is determined by majority vote of the base CNN and the FCSNs. Experimental results using the MNIST, Fashion-MNIST and CIFAR-10 datasets show that the proposed approach further improves the performance of CNNs. In particular, an EnsNet achieves a state-of-the-art error rate of 0.16% on MNIST.},
keywords={},
doi={10.1587/transinf.2022EDL8098},
ISSN={1745-1361},
month={July},}
Copy
TY - JOUR
TI - Ensemble Learning in CNN Augmented with Fully Connected Subnetworks
T2 - IEICE TRANSACTIONS on Information
SP - 1258
EP - 1261
AU - Daiki HIRATA
AU - Norikazu TAKAHASHI
PY - 2023
DO - 10.1587/transinf.2022EDL8098
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E106-D
IS - 7
JA - IEICE TRANSACTIONS on Information
Y1 - July 2023
AB - Convolutional Neural Networks (CNNs) have shown remarkable performance in image recognition tasks. In this letter, we propose a new CNN model called the EnsNet which is composed of one base CNN and multiple Fully Connected SubNetworks (FCSNs). In this model, the set of feature maps generated by the last convolutional layer in the base CNN is divided along channels into disjoint subsets, and these subsets are assigned to the FCSNs. Each of the FCSNs is trained independent of others so that it can predict the class label of each feature map in the subset assigned to it. The output of the overall model is determined by majority vote of the base CNN and the FCSNs. Experimental results using the MNIST, Fashion-MNIST and CIFAR-10 datasets show that the proposed approach further improves the performance of CNNs. In particular, an EnsNet achieves a state-of-the-art error rate of 0.16% on MNIST.
ER -