I attempt to share some learnings from Tsai et al. (2019) connecting the Attention Mechanism to Nadaraya-Watson Kernel Smoothing (1964).

1

I attempt to construct a geometric understanding of how Self-Attention works in Large Language Models.

2

I attempt to explain the concept of Gaussian Copulas as simply as possible.

3

I try to construct a geometric intuition for why Hessian matrices must be Positive Semi-Definite in convex optimization.

4

I attempt to construct a geometric understanding of how SVD works, by building upon previous posts about spectral decomposition and various transformation matrices.

5

I attempt to construct a geometric understanding of how eigendecomposition works.

6

I write about (what I think is) an intuitive way to learn about the Fourier Transform, and share some excellent resources.

7

I write about some essential matrix transformations that help build an understanding of matrix multiplication as a composition of matrix transformations.

8

I write about the kernel trick and show an informal derivation when applied to a regularized least squares example

9

I give an overview of how basis functions are used in linear regression models. Then, I derive the solution to an OLS problem with basis functions involved.

10

I share my revision on basic linear algebra, and attempt to construct a geometric intuition based on 3B1B's teaching material.

11

I share some context for the blog here.

12