The underlying mechanisms of alignment in error back-propagation through arbitrary weights
Alireza Rahmansetayesh, Ali Ghazizadeh, Farokh Marvasti
This paper investigates the mechanisms of weight alignment in BP-TAW, showing it is governed by statistical properties of neural signals and can be improved with weight normalization.
Neurocomputing · Sep 23, 2024Peer-reviewedNeuroscience