A novel continual learning framework, Ensemble of Expert Embedders (E3), enables accurate detection of images from newly emerged synthetic image generators using minimal training data.
