PEPITA – Giorgia Dellaferrera et al. – Open Neuromorphic
Training deep neural networks relies on backpropagation, which lacks biological plausibility. PEPITA is an alternative learning rule that trains DNNs by replacing the backward pass of the backpropagation algorithm with a second forward pass.
In this talk, Giorgia Dellaferrera and the other authors present the ideas behind the algorithm, its development and implementation in PyTorch.
