Machine Unlearning for Source-Free Unsupervised Partial-Domain Adaptation in Remote Sensing

Source-free unsupervised domain adaptation (SFUDA) enables the model adaptation to unlabeled target domains without accessing source data. However, when the source domain contains classes absent in the target domain, existing methods suffer from negative transfer (NT): knowledge of irrelevant source-only classes interferes with target class recognition, significantly degrading classification accuracy. We propose machine unlearning-based SFUDA (MUSFUDA), which addresses this problem by selectively unlearning source-only class knowledge from the pretrained model rather than adding compensatory mechanisms. This machine unlearning approach allows the model to focus on shared classes, fundamentally eliminating NT. Remote sensing images with large intraclass variations and high interclass similarity cause over-unlearning of target classes when forgetting source-only classes; thus, we design the model-disruption-based dual-teacher unlearning strategy (MDUS), which uses dual teachers to manage target class preservation and source-only class erasure through knowledge distillation. MDUS is lightweight and easily integrated into existing SFUDA frameworks. Experiments on remote sensing datasets demonstrate that combining MDUS with representative baselines consistently reduces NT and improves classification performance, maintaining high efficiency, validating the effectiveness and generalizability of our approach.