Abstract: Existing Cross-Domain Few-Shot Learning (CDFSL) methods require access to source domain data to train a model in the pre-training phase. However, due to increasing concerns about data ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results