arXiv Open Access 2019

Constrained Linear Data-feature Mapping for Image Classification

Juncai He Yuyan Chen Lian Zhang Jinchao Xu

Lihat Sumber

Abstrak

In this paper, we propose a constrained linear data-feature mapping model as an interpretable mathematical model for image classification using convolutional neural network (CNN) such as the ResNet. From this viewpoint, we establish the detailed connections in a technical level between the traditional iterative schemes for constrained linear system and the architecture for the basic blocks of ResNet. Under these connections, we propose some natural modifications of ResNet type models which will have less parameters but still maintain almost the same accuracy as these corresponding original models. Some numerical experiments are shown to demonstrate the validity of this constrained learning data-feature mapping assumption.

Topik & Kata Kunci

eess.IV cs.CV math.NA

Penulis (4)

Juncai He

Yuyan Chen

Lian Zhang

Jinchao Xu

Format Sitasi

APA MLA BibTeX

He, J., Chen, Y., Zhang, L., Xu, J. (2019). Constrained Linear Data-feature Mapping for Image Classification. https://arxiv.org/abs/1911.10428

Akses Cepat

Lihat di Sumber

Informasi Jurnal

Tahun Terbit: 2019
Bahasa: en
Sumber Database: arXiv
Akses: Open Access ✓