Softmax Cross Entropy Loss with Unbiased Decision Boundary for Image Classification

CORC > 兰州理工大学 > 兰州理工大学

	Softmax Cross Entropy Loss with Unbiased Decision Boundary for Image Classification
	Cao, Jie 1; Su, Zhe 1; Yu, Liyun 1; Chang, Dongliang 1; Li, Xiaoxu 1; Ma, Zhanyu 2
	2018
关键词	image classification neural networks softmax cross entropy loss decision boundary
页码	2028-2032
英文摘要	Considering that in neural network based on softmax cross entropy loss, the output probability is mainly based on linear computation of parameter vectors of each class in the last layer and hidden features in the layer of sample points. Therefore, the final output of neural network is effected by of the L2-norm of parameter vector of each class. Taking binary-class as an example, if the parameter vector of a class has a large L2-norm, decision boundary is close to another class with smaller L2-norm, so that sample points will be easily assigned to the class with large L2-norm. Based on it, this paper proposes a new softmax cross entropy loss, which adjusts the position of decision boundary so that it is not biased to any class. Experimental results on the LabelMe dataset and the UIUC-Sports dataset show that the proposed loss is superior to softmax cross entropy loss.
会议录	2018 CHINESE AUTOMATION CONGRESS (CAC)
会议录出版者	IEEE
会议录出版地	345 E 47TH ST, NEW YORK, NY 10017 USA
语种	英语
WOS研究方向	Automation & Control Systems
WOS记录号	WOS:000459239502016
内容类型	会议论文
源URL	[http://119.78.100.223/handle/2XXMBERH/36192]
专题	兰州理工大学
通讯作者	Cao, Jie
作者单位	1.Lanzhou Univ Technol, Sch Comp & Commun, Lanzhou, Gansu, Peoples R China 2.Beijing Univ Posts & Telecommun, Sch Informat & Commun, Beijing, Peoples R China
推荐引用方式 GB/T 7714	Cao, Jie,Su, Zhe,Yu, Liyun,et al. Softmax Cross Entropy Loss with Unbiased Decision Boundary for Image Classification[C]. 见:.