OmnıLearn
Machine Learning: the fundamentals
Text· 12 minMembers-only lesson

Measuring and fixing error: loss and gradient descent

A model needs a way to measure how wrong its predictions are, so it can later improve them. This measure is called loss. The lower the loss, the closer the model matches the real data. Two common ways to measure…

access denied :: authentication required

The rest is for members

Create an account to open the full course, save your progress and take the graded quizzes.

The first 2 lessons of this course are open to everyone.

We measure site traffic (GA4) and, if you say yes, we also tune Meta ads. Nothing is stored on your device before you answer, and your choice sticks for 6 months.