An unsupervised domain adaptation method for distinguishing simple and difficult samples

Abstract

The application discloses an unsupervised domain adaptation method for distinguishing simple samples from difficult samples, comprising the following steps: distinguishing target domain samples according to the entropy values of the target domain samples, defining samples with entropy values greater than or equal to a preset entropy threshold value as simple samples, and defining samples with entropy values less than the preset entropy threshold value as difficult samples; for the target domain samples classified as simple samples, a well-trained classifier of a source domain is used to assign pseudo labels to the target domain samples; for the target domain samples classified as difficult samples, the unsupervised domain adaptation is adjusted to semi-supervised domain adaptation by using the simple samples with assigned pseudo labels in step S2, a more robust classifier is trained by using the source domain labels and the labels of the simple samples of the target domain, the class centers are calculated, and the inter-domain contrast alignment and the instance contrast alignment are respectively optimized to reduce the inter-domain and intra-domain differences. The application can solve the problem of classification errors of difficult samples in the target domain in the existing domain adaptation method.

Images