For feature fusion, should features come from seen or unseen data? If seen, how do we avoid bias or leakage when training the final classifier?

In feature fusion, should we extract features from the same (seen) data used to train the base models, or from unseen data? Since these features are later used to train a final classifier, I’m...

05 April 2025 4,240 0 View