Title: Neural Networks through the lens of Category Theory
Abstract: I will give an introduction to the categorical foundation of gradient-based learning algorithms. I’ll show how category theory can be used to formalize various components of neural networks: parameterization, bidirectionality and differentiation. I’ll also show how CT can be used as a graphical language and a visual, pedagogical aid in understanding neural networks. Time permitting, we will cover how abstract categorical results of leanring can be transferred to the surprising setting of Boolean circuits.