Skip to content
Truth Academy.
Truth Academy
← All modules
The Math UnderneathAdvanced45 min

Continuous Mathematical Methods for ML

The numerical math that makes learning actually compute: linear systems, derivatives, and the stability of it all.

AI tutor is turned off for this class

Use the CS205L lectures, notes, and assignments below to keep learning.

Big Idea

The Math Underneath

Grade bands

K-2 · 3-5 · 6-8 · 9-12

View on the ladder →

Lesson overview

The numerical math that makes learning actually compute: linear systems, derivatives, and the stability of it all.

Teacher script · 45 min

  1. 0–5

    Hook

    On paper you write '1/3' exactly. A computer stores 0.33333… and rounds. Do millions of such steps and tiny rounding errors can snowball or cancel. This field studies how to do continuous math (calculus, linear algebra) on machines that can only approximate, without the answer falling apart.

  2. 5–15

    Explore

    Students do the activity in pairs: Solve two equations with two unknowns by elimination. Now imagine a million of them.

  3. 15–30

    Explain

    Some problems amplify errors: a hair's-width change in input swings the output wildly (ill-conditioned). Some algorithms add error of their own (unstable). Reliable numerical computing means understanding both, so your trained model doesn't silently produce garbage. This is why ML practitioners normalize data and watch for exploding gradients.

  4. 30–40

    Connect to the summit

    Read aloud the "From the summit" section below. Ask students which ideas from today they recognize in it.

  5. 40–45

    Check

    Run the formative check below. Anyone who can explain a key term in their own words has it.

Student activity

Solve two equations with two unknowns by elimination. Now imagine a million of them.

Slides

1Title: Continuous Mathematical Methods for ML
2Hook: Math the computer can really do
3Do it: Solving many equations at once
4How it works: Conditioning and stability
5Key idea: Floating point
6Key idea: Conditioning
7Key idea: Numerical stability
8From the summit: the Stanford course

Formative check

  • 1.In your own words, what is "Floating point"? (Looking for: How computers store real numbers approximately, introducing tiny rounding errors.)
  • 2.In your own words, what is "Conditioning"? (Looking for: How much a problem amplifies small input errors into large output errors.)
  • 3.In your own words, what is "Numerical stability"? (Looking for: Whether an algorithm keeps its own rounding errors from blowing up.)

Carry-away concepts

Floating point
How computers store real numbers approximately, introducing tiny rounding errors.
Conditioning
How much a problem amplifies small input errors into large output errors.
Numerical stability
Whether an algorithm keeps its own rounding errors from blowing up.
Least squares
Finding the best-fit answer when you have more equations than unknowns.

From the summit · the Stanford source

You study the continuous-math engine of ML: numerical linear algebra, optimization, conditioning, and why algorithms break on real hardware.

This module is based on a Stanford course.