From the 1 of 1 linked paper with an AI index.
1 paper
Junlong Shen, Xingyu Li
The paper shows that for common deep network layers with activation, normalization, and linear maps, the Jacobian‑transpose needed for predictive coding can be expressed using only…