Search papers, labs, and topics across Lattice.
1
0
2
Focal loss doesn't just re-weight examples; it fundamentally reshapes the learned probability distribution, potentially distorting it far from the true data distribution, especially with severe class imbalance.
