I wrote a blog post! In an ML paper reading group awhile back, we were going through a paper on flow models and I realized I didn't really understand the change-of-variables formula in probability as well as I should. So I wrote up this post and put together some interactive visuals to help me more fully understand my main point of confusion (how/why the Jacobian inverse term plays into the formula).
I'm sharing my writeup here in the hope that folks will find the visualizations as useful as I did. I hope you enjoy reading, and I'm always grateful for feedback on the prose or visuals if you have any!
I'm sharing my writeup here in the hope that folks will find the visualizations as useful as I did. I hope you enjoy reading, and I'm always grateful for feedback on the prose or visuals if you have any!
reply