期刊论文

Haofeng Zhang

英文

Information Sciences

0020-0255

2023

625

536-552

10.1016/j.ins.2023.01.039

Suppose T is a dataset, Ts,Tu are two subsets of it and satisfy T=Ts∪Tu,Ts∩Tu=∅, where Ts represents the set of seen class samples, and Tu stands for the set of unseen class samples. There are ns smples of c categories in Ts={xi,ayi|xi∈Rd*ns,ayi∈Rk*c},i=1,2,…,ns, where i represents the sample index in the seen class, and xi represents the ith sample’s d-dimensional feature vector, and yi represents the category of the i sample, and ayi represents the k-dimensional semantic feature of the yi

本校为其他机构

Traditional Generative Adversarial Network (GAN) based Generalized Zero Shot Learning (GZSL) methods usually suffer from a problem that these methods ignore the differences between classes when using the standard normal distribution to fit the true distribution of each category, and the incompleteness of a single adversarial training makes the model unable to capture all the characteristics of the samples. To address this problem, a data-driven recurrent adversarial generative network is proposed in this paper. We first synthe-size visual prototypes for unseen classes using the transformation ...