As such, it is often desired to select a small number of informative features to consider in subsequent analyses.

Fairness in machine learning has attracted increasing attention in recent years.

The meta generator is termed as MGDD in our approach. Once adapted, it can handle arbitrary sizes of synthetic datasets, even for those unseen during adaptation.

Neural Information Processing Systems http://nips.cc/

To better reuse the VLM resource and fully leverage its potential on different zero-shot image classification tasks, a promising strategy is selecting appropriate Pre-Trained VLMs from the VLM Zoo, relying solely on the text data of the target dataset without access to the dataset's images.