Recent content on David Gurevich
I’ve honestly spent a lot of time thinking about outliers and change detection in real-time data streams. You can get into some pretty serious rabbit holes when thinking about this stuff. If you’re not careful, you might accidentally end up in non-parametric land with quickest change detection or some other crazy ideas1. I think something a bit more down-to-earth2 and maybe more…
When I started writing this post, I was halfway through writing what’s basically an introductory article to non-parametric anomaly detection in a data stream. I figure that I should write about more pressing matters, however, and so that’s why I’m writing this post about buying sweaters. What’s so hard about buying a sweater anyways? This is a great question. Completely…
Image inpainting is the task of restoring missing or corrupted parts of an image. State of the art methods are largely generative methods that are pretrained on hundreds of thousands of images, and attempt to sample from the latent distribution of the image. Rather than taking the “deep” approach, we will explore methods that are entirely based on self-similarity measures. The methods…
One of my favourite courses from my time at the University of Waterloo was AMATH 391. At the time, it had the provocative title From Fourier to Wavelets. Nowadays, it has the more sober name Data Analysis with Fourier and Wavelet Methods. Honestly, this is probably a better name for the course anyways. I want to present to you a somewhat simple concept from this course. When we take the Fourier…