BLOG
note: website is under construction
[POST IN PROGRESS]. In machine learning, where empirical studies rule and explainability is literally an entire subfield, provable theoretical guarantees don’t come around often. When I first learned about dimensionality reduction methods, I was impressed by Isomap for various reasons, not least its “[guarantee] asymptotically to recover the true dimensionality and geometric structure of a strictly larger class of nonlinear manifolds.” There are plenty of casual articles describing how Isomap works, but the proof of said guarantee is usually glossed over (and omitted from the main body of the original paper), so I wanted to write the proof down for myself here.
a gazillion ways to think about variable selection, rethinking linear algebra, an overview of discrete combinatorial geometric theorems, coarse notions of curvature on graphs and graph learning
building a note-taking site from scratch isn’t exactly ideal; there are lots of great note-taking applications out there. i’m interested in adding tools that, to my knowledge, don’t exist outside of their own apps–many of which i have individual gripes with–or are actually nonexistent.
i experimented with different personal websites before settling on neocities. i first tried notion, which is a bit like publishing a giant google spreadsheet with links to all your notes. i didn’t love the lack of customization (beyond choosing different… emojis…?), either. github pages was also an obvious option; though i made an academic-ish profile, it felt sterile and i never felt like i had anything to post.