Math for ML
You grasp, hands-on, the math behind attention, embeddings, and LoRA.
- Who it's for
- adults entering ML
- Format
- one-on-one or mini-group, online
- Level
- a bridge from school math to ML
What you'll learn
- You understand vectors and matrices as transformations.
- You derive backprop for a two-layer net by hand.
- You apply probability and statistics in models.
- You code gradient descent yourself.
Program by modules
- 01
Linear algebra
- Vectors, matrices as transformations
- Eigenvalues, SVD
Project: Compress an image with SVD.
- 02
Calculus
- Derivatives, gradients
- Chain rule, backprop
Project: Derive backprop for a two-layer net by hand.
- 03
Probability & statistics
- Distributions, Bayes
- MLE, expectation and variance
Project: Compute an MLE from data and verify it.
- 04
Optimization
- Gradient descent, momentum
- Adam, learning rate
Project: Compare optimizer convergence.
- 05
From math to code
- Implementing in numpy
- From formula to working code
Project: Write and train linear regression from scratch.