Machine Learning Seminar Series

Location: ED 612

Speaker: Affan Shoukat, University of Regina

Title: Deep-dive into the backpropagation algorithm

Abstract:

The backpropagation algorithm is an iterative method used to train neural networks by calculating gradients of the loss function with respect to the network parameters. It is a two-step process.  First, a forward pass over the network predicts the outputs and calculates the loss, followed by a backward pass to update the weights based on computed gradients.