Unsupervised Domain Expansion from Multiple Sources
This addresses domain expansion for machine learning systems needing adaptation to new data without retraining on all previous domains, though it appears incremental as it builds on existing domain adaptation concepts.
The paper tackles the problem of adapting a system to new domains without forgetting previous ones, known as domain expansion, by proposing an unsupervised multi-source method that uses predicted class probabilities to mitigate biases and preserve performance, achieving verified effectiveness on datasets like VLCS, ImageCLEF_DA, and PACS.
Given an existing system learned from previous source domains, it is desirable to adapt the system to new domains without accessing and forgetting all the previous domains in some applications. This problem is known as domain expansion. Unlike traditional domain adaptation in which the target domain is the domain defined by new data, in domain expansion the target domain is formed jointly by the source domains and the new domain (hence, domain expansion) and the label function to be learned must work for the expanded domain. Specifically, this paper presents a method for unsupervised multi-source domain expansion (UMSDE) where only the pre-learned models of the source domains and unlabelled new domain data are available. We propose to use the predicted class probability of the unlabelled data in the new domain produced by different source models to jointly mitigate the biases among domains, exploit the discriminative information in the new domain, and preserve the performance in the source domains. Experimental results on the VLCS, ImageCLEF_DA and PACS datasets have verified the effectiveness of the proposed method.