by Maheen Rassell

(No PyTorch or TensorFlow Just Linear Algebra!)

Why: Getting a Deeper Understanding

I am very interested in AI / ML but everyone knows it at the surface level

Untitled

we connect layers and nodes and use complex models makes cool predictions

but kind of unclear how it REALLY works…

cant get much closer unless you implement it from scratch and really learn it from the ground up!

Credit goes to Samson Zhang’s video where I learned the math:

https://www.youtube.com/watch?v=w8yWXqWQYmU&ab_channel=SamsonZhang

Personal Demo:

https://www.youtube.com/watch?v=RmDJ16-DuYU&ab_channel=MaheenRassell

Problem Statement:

The MNIST (Modified National Institute of Standards and Technology) dataset is a large collection of handwritten digits. It consists of:

  1. 60,000 training images
  2. 10,000 testing images
  3. Each image is a 28x28 pixel grayscale image