TY - CHAP
T1 - Convolutional Neural Network Ensemble Fine-Tuning for Extended Transfer Learning
AU - Korzh, Oxana
AU - Joaristi, Mikel
AU - Serra, Edoardo
N1 - Publisher Copyright:
© Springer International Publishing AG, part of Springer Nature 2018.
PY - 2018/1/1
Y1 - 2018/1/1
N2 - Nowadays, image classification is a core task for many high impact applications such as object recognition, self-driving cars, national security (border monitoring, assault detection), safety (fire detection, distracted driving), geo-monitoring (cloud, rock and crop-disease detection). Convolutional Neural Networks(CNNs) are effective for those applications. However, they need to be trained with a huge number of examples and a consequently huge training time. Unfortunately, when the training set is not big enough and when re-train the model several times is needed, a common approach is to adopt a transfer learning procedure. Transfer learning procedures use networks already pretrained in other context and extract features from them or retrain them with a small dataset related to the specific application (fine-tuning). We propose to fine-tuning an ensemble of models combined together from multiple pretrained CNNs (AlexNet, VGG19 and GoogleNet). We test our approach on three different benchmark datasets: Yahoo! Shopping Shoe Image Content, UC Merced Land Use Dataset, and Caltech-UCSD Birds-200-2011 Dataset. Each one represents a different application. Our suggested approach always improves accuracy over the state of the art solutions and accuracy obtained by the returning of a single CNN. In the best case, we moved from accuracy of 70.5% to 93.14%.
AB - Nowadays, image classification is a core task for many high impact applications such as object recognition, self-driving cars, national security (border monitoring, assault detection), safety (fire detection, distracted driving), geo-monitoring (cloud, rock and crop-disease detection). Convolutional Neural Networks(CNNs) are effective for those applications. However, they need to be trained with a huge number of examples and a consequently huge training time. Unfortunately, when the training set is not big enough and when re-train the model several times is needed, a common approach is to adopt a transfer learning procedure. Transfer learning procedures use networks already pretrained in other context and extract features from them or retrain them with a small dataset related to the specific application (fine-tuning). We propose to fine-tuning an ensemble of models combined together from multiple pretrained CNNs (AlexNet, VGG19 and GoogleNet). We test our approach on three different benchmark datasets: Yahoo! Shopping Shoe Image Content, UC Merced Land Use Dataset, and Caltech-UCSD Birds-200-2011 Dataset. Each one represents a different application. Our suggested approach always improves accuracy over the state of the art solutions and accuracy obtained by the returning of a single CNN. In the best case, we moved from accuracy of 70.5% to 93.14%.
KW - CNN
KW - deep learning
KW - image classification
KW - transfer learning
UR - https://scholarworks.boisestate.edu/cs_facpubs/160
UR - https://doi.org/10.1007/978-3-319-94301-5_9
UR - http://www.scopus.com/inward/record.url?scp=85049370841&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-94301-5_9
DO - 10.1007/978-3-319-94301-5_9
M3 - Chapter
SN - 9783319943008
T3 - 0302-9743
SP - 110
EP - 123
BT - BigData 2018
A2 - Khan, Latifur
A2 - Zhang, Liang-Jie
A2 - Lee, Kisung
A2 - Chin, Francis Y.
A2 - Chen, C. L.
PB - Springer Verlag
T2 - 7th International Congress on Big Data, BigData 2018 Held as Part of the Services Conference Federation, SCF 2018
Y2 - 25 June 2018 through 30 June 2018
ER -