
The aim of this paper is to introduce a new learning procedure for neural networks and to demonstrate that it works well enough on a few small problems to be worth further investigation. The Forward-Forward algorithm replaces the forward and backward passes of backpropagation by two forward passes. 2022: G. Hinton https://arxiv.org/pdf/2212.13345v1.pdf
Version: 20241125
Comments (0)
To leave or reply to comments, please download free Podbean or
No Comments
To leave or reply to comments,
please download free Podbean App.