Enhance the Efficacy of Deep CNN with Auxiliary Labels
Authors: 
Fen Fang, Liyuan Li, Joo-Hwee LimAuthors Info & Claims
Fen Fang, Liyuan Li, Joo-Hwee LimAuthors Info & ClaimsPages 76 - 81
Abstract
Auxiliary attributes can improve the performance of deep CNNs for image classification. However, not all auxiliary tasks are helpful for main task performance. Therefore, in this paper we focus on improving the efficacy of deep CNN models with the support of auxiliary attributes. On the principle of minimizing cross entropy loss, we derive a new algorithm to iteratively train the deep CNN model on target and auxiliary labels. We demonstrate that by introducing auxiliary attributes to training images, uncertainty can be reduced in target classification tasks, and adversarial effects avoided in multi-task formulation. We evaluated our learning approach on three categories of overlapping image sets for identical, partial overlapping, and disjoint situations. We performed three group of experiments on three popular deep CNN networks, and three challenging datasets. The results show that our method is able to improve efficacy for target tasks with auxiliary labels in situations where the multi-task learning fails, or is not applicable.
References
[1]
Alonso, H. and Plank, B. 2017. When is multitask learning effective? Semantic sequence prediction under varying data conditions. In EACL.
[2]
Bjerva, J. 2017. Will my auxiliary tagging task help? Estimating auxiliary tasks effectivity in multi-task learning. In NCCL. 216--220.
[3]
Kendall, A., Gal, Y., and Cipolla, R. 2018. Multi-task learning using uncertainty to weigh losses for scene geometry and semantics. In CVPR.
[4]
Liebel, L. and Körner, M. 2018. Auxiliary tasks in multi-task learning. CoRR, abs/1805.06334.
[5]
Liang, K., Guo, Y., Chang, H., and Chen, X. 2017. Incomplete attribute learning with auxiliary labels. In IJCAI. 2252--2258.
[6]
Russakovsky, O. and Fei-Fei, L. 2010. Attribute learning in large-scale datasets. In ECCV Workshop.
[7]
Romera-Paredes, B., Argyriou, A., Berthouze, N., and Pontil, M. 2012. Exploiting unrelated tasks in multi-task learning. In ICAIS. 951--959.
[8]
Ruder, S. 2017. An overview of multi-task learning in deep neural networks. arXiv:1706.05098.
[9]
Zhang, Z., Luo, P., Loy, C., and Tang, X. 2016. Learning deep representation for faced alignment with auxiliary attributes. TPAMI, 38, 918--930
[10]
Huh, M.-Y., Agrawal, P., and Efros, A. A. 2016. What makes imagenet good for transfer learning? CoRR, abs/1608.08614.
[11]
Khan, S., Hayat, M., and Porikli, F. 2017. Scene categorization with spectral features. In ICCV.
[12]
Hinton, G., Vinyals, O., and Dean, J. 2015. Distilling the knowledge in a neural network. NIPS Deep Learning and Representation Learning Workshop.
[13]
Pereyra, G., Tucker, G., Chorowski, J., Kaiser, L., and Hinton, G. 2017. Regularizing neural networks by penalizing confient output distributions. In ICLR Workshop.
[14]
Jia, Y., Shelhamer, E., Donahue, J., Karayev, S., Long, J., Girshick, R., Guadarrama, S., and Darrell, T. 2014. Caffe: Convolutional architecture for fast feature embedding. ACM MM.
[15]
Krause, J., Sapp, B., Howard, A., Zhou, H., Toshev, A., Duerig, T., Philbin, J., and Fei-Fei, L. 2016. The unreasonable effectiveness of noisy data for fine-grained recognition. In ECCV.
[16]
Krause, J., Stark, M., Deng, J., and Fei-Fei, L. 2013. 3d object representations for fine-grained categorization. In ICCV Workshops. 554--561.
[17]
Zheng, H., Fu, J., Mei, T., and Luo, J. 2017. Learning multi-attention convolutional neural network for fine-grained image recognition. In ICCV.
[18]
Khosla, A., Jayadevaprakash, N., Yao, B., and Fei-Fei, L. 2011. Novel dataset for fine-grained image categorization. In CVPR.
[19]
Zhang, L., Yang, F., Zhang, Y. D., and Zhu, Y. J., 2016. Road crack detection using deep convolutional neural network. In ICIP.
[20]
Xie, S., Yang, T., Wang, X., and Lin, Y. 2015. Hyper-class augmented and regularized deep learning for fine-grained image classification. In CVPR.
[21]
Krizhevsky, A., Sutskever, I., and Hinton, G. 2012. Imagenet classification with deep convolutional neural networks. In NIPS.
[22]
Simonyan, K. and Zisserman, A. 2014. Very deep convolutional networks for large-scale image recognition. In CVPR.
[23]
Zhou, F. and Lin, Y. 2016. Fine-grained image classification by exploring bipartite-graph labels. In CVPR.
[24]
Hou, S., Liu, X., and Wang, Z. 2017. Dualnet: Learn complementary features for image recognition. ICCV.
[25]
Hu, Y. and Zhao, C. 2010. A local binary pattern based methods for pavement crack detection. Journal of Pattern Recognition Research. 5, 140--147.
[26]
Kapela, R., Sniataa, P., Turkot, A., Rybarczyk, A., Pozarycki, A., Rydzewski, P., Wyczaek, M., and Bloch, A. 2015. Asphalt surfaced pavement cracks detection based on histograms of oriented gradients. Int. Conf. on Mixed Design of Integrated Circuits and Systems. 579--584.
[27]
Oliveira, H. and Correia, P. 2009. Automatic road crack segmentation using entropy and image dynamic thresholding. In Eur Conference on Signal Process. 622--626.
[28]
Salman, M., Mathavan, S., Kamal, K., and Rahman, M. 2013. Pavement crack detection using the gabor filter. In International Conference on Intelligent Transportation Systems. 2039--2044. IEEE.
[29]
van-der Maaten, L. and Hinton, G. 2008. Visualizing data using t-sne. Journal of Machine Learning Research. 9, 2579--2605.
Index Terms
- Enhance the Efficacy of Deep CNN with Auxiliary Labels
Recommendations
Using CNN for Encoder Optimization in H.265/HEVC
MOBIMEDIA'17: Proceedings of the 10th EAI International Conference on Mobile Multimedia CommunicationsIn this work-in-progress paper, we proposed using deep learning techniques, especially the deep Convolutional Neural Network (CNN) to perform critical tasks of video ending within the framework of H.265/HEVC. Deep CNNs have achieved break-through ...
Deep Residual 1-Dimensional Convolutional Neural Networks in Vision
Pattern RecognitionAbstractWhile 2D convolutional neural networks (CNNs) demonstrate outstanding performance on computer vision tasks, their computational costs remain high. This paper reduces computational costs by introducing a novel architecture that replaces spatial 2D ...
Improved softmax loss for deep learning‐based face and expression recognition
In recent years, deep convolutional neural networks (CNN) have been widely used in computer vision and significantly improved the performance of image recognition tasks. Most works use softmax loss to supervise the training of CNN and then adopt the ...
Comments
Information & Contributors
Information
Published In
November 2019
108 pages
ISBN:9781450372350
DOI:10.1145/3373724
Copyright © 2019 ACM.
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from [email protected]
Publisher
Association for Computing Machinery
New York, NY, United States
Publication History
Published: 03 February 2020
Check for updates
Author Tags
Qualifiers
- Research-article
- Research
- Refereed limited
Conference
ICRAI '19
ICRAI '19: 2019 5th International Conference on Robotics and Artificial Intelligence
November 22 - 24, 2019
Singapore, Singapore
Contributors
Other Metrics
Bibliometrics & Citations
Bibliometrics
Article Metrics
- 0Total Citations
- 143Total Downloads
- Downloads (Last 12 months)7
- Downloads (Last 6 weeks)0
Reflects downloads up to 03 Mar 2025
Other Metrics
Citations
View Options
Login options
Check if you have access through your login credentials or your institution to get full access on this article.
Sign in