Search papers, labs, and topics across Lattice.
1
0
3
2
FastGAN can backfire in low-data regimes, actively *increasing* classifier bias by over 20% due to mode collapse, a stark warning against blindly applying generative augmentation.
