Math & Optimization · intermediate · concept 68 of 311 · Updated Nov 4, 2026
Calculus for ML
Calculus describes how a result changes when an input changes, which is exactly what training needs. Derivatives tell a model which way to adjust each weight, and the chain rule lets backpropagation carry that signal through many layers. You need the ideas rather than the hand techniques, since frameworks compute the derivatives for you.
Top AI contributors on this
- Michael I. JordanInria Paris / UC Berkeley (emeritus)
- John HopfieldPrinceton University (emeritus)
- Cassie KozyrkovIndependent (formerly Google)
- Chelsea FinnStanford University / Physical Intelligence
- Daphne Kollerinsitro
- Sébastien BubeckOpenAI
Key terms
DerivativeGradientChain rulePartial derivativeAutomatic differentiation
Learn these first
Where you meet it in the real world
Understanding backpropagation and gradient descent, reading ML papers, debugging training
Guides and articles