ABSTRACT
In the computer vision field, there are many problems to be solved such as consideration of connectivity between problems. These concerns are still indispensable. Existing approaches usually proceed with transfer learning of all problems through one space. This can be very complex and inefficient for one model to solve, and if the data does not have a constant distribution, it causes difficulties in extracting explicit features in multi-classification. In this paper, we propose a method for independent divided learning motivated by the divide and conquer algorithm. Our approach split the multi-classification problem into multiple small units so that all problems can be solved with multiple learners rather than all at once. In addition, it is not randomly dividing the data, but based on the similarity matrix recognized by hierarchical clustering, so that each divided learner can extract explicit features. As a result, the classifier brings a more balanced and improved performance than general transfer learning.
- Fernando, B., Habrard, A., Sebban, M., & Tuytelaars, T. (2013). Unsupervised visual domain adaptation using subspace alignment. In Proceedings of the IEEE international conference on computer vision (pp. 2960--2967).Google ScholarDigital Library
- Gong, B., Shi, Y., Sha, F., & Grauman, K. (2012, June). Geodesic flow kernel for unsupervised domain adaptation. In 2012 IEEE conference on computer vision and pattern recognition (pp. 2066--2073). IEEEGoogle ScholarCross Ref
- Zelikin, M. I. (2000). Control Theory and Optimization I. Encyclopedia of Mathematical Sciences, vol. 86.Google ScholarCross Ref
- Pan, S. J., Ni, X., Sun, J. T., Yang, Q., & Chen, Z. (2010, April). Cross-domain sentiment classification via spectral feature alignment. In Proceedings of the 19th international conference on World wide web (pp. 751--760).Google ScholarDigital Library
- Krizhevsky, A., & Hinton, G. (2009). Learning multiple layers of features from tiny images.Google Scholar
- He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 770--778).Google ScholarCross Ref
- Kolesnikov, A., Beyer, L., Zhai, X., Puigcerver, J., Yung, J., Gelly, S., & Houlsby, N. (2020). Big transfer (bit): General visual representation learning. In Computer Vision-ECCV 2020: 16th European Conference, Glasgow, UK, August 23--28, 2020, Proceedings, Part V 16 (pp. 491--507). Springer International Publishing.Google Scholar
- Simonyan, K., & Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556.Google Scholar
- Van der Maaten, L., & Hinton, G. (2008). Visualizing data using t-SNE. Journal of machine learning research, 9(11).Google Scholar
Index Terms
- A similarity-based deep feature extraction method using divide and conquer for image classification
Recommendations
Hierarchical image feature extraction and classification
MM '10: Proceedings of the 18th ACM international conference on MultimediaIn the field of machine learning and pattern recognition, an alternative to conventional classification is hierarchical classification that exploits hierarchical relations between concepts of interest. To the best of our knowledge, all hierarchical ...
Genetic programming for feature extraction and construction in image classification
AbstractGenetic Programming (GP) has been successfully applied to image classification and achieved promising results. However, most existing methods either address binary image classification tasks only or need a predefined classifier to ...
Highlights- A new GP representation to get an effective combination of features and classifiers.
Multi-view Multi-task Feature Extraction for Web Image Classification
MM '14: Proceedings of the 22nd ACM international conference on MultimediaThe features used in many multimedia analysis-based applications are frequently of very high dimension. Feature extraction offers several advantages in highly dimensional cases, and many recent studies have used multi-task feature extraction approaches, ...
Comments