--

Hi, you can read the para right after Eq 6 in the original paper to see how authors come up wit this. But short answer is, X_i gets absorbed into b_i and b_k is added to restore exchange symmetry in the Equation.

--

--

Thushan Ganegedara
Thushan Ganegedara

Written by Thushan Ganegedara

Google Dev Expert | ML @ Canva | Author📗| PhD👨🏾‍🎓. Youtube: bit.ly/deeplearninghero | LinkedIN: thushan.ganegedara | Medium: bit.ly/support-thushan

Responses (1)