Early Stopping and its Faults
Introduction Seeing as how the last few posts have been a bit theoretical in nature, I thought it might be useful to switch to something more practical. Early stopping is a strategy...
Hello, I’m Alex Adam
I'm a PhD researcher at the University of Toronto focused on making the deployment of deep learning models safe, robust, and reliable.
Latest Posts
Universal Approximation Theorem - Part 3
Introduction It’s been too long since my last post, so expect to see a chain of new posts in the coming months. Back to the UAT madness that I started looking at last year. In part...
Universal Approximation Theorem - Part 2
Introduction In today’s post, we look at fitting a quartic function. Furthermore, we’re going to take a look at how far away we have to move from our training dataset in order for o...
Universal Approximation Theorem - Part 1
Introduction Early on in my deep learning studies, I heard something along the lines of: neural networks are universal function approximators. This made me really excited about the ...