Syllabus Lesson 124 of 246 · Module 10: Networks that see (CNNs)
Module 10: Networks that see (CNNs)

Gate: a CNN for a new image task (AI off)

Why it matters Designing a network under a budget, alone, is what this module was for.

AI tutor off. New shape images, and a budget: at most 3,000 weights in total. Use any of the house layers.

Your turn

Write build_and_train(X, y) returning (predict, params), where params is the list of every weight Tensor. The check counts every Tensor your predict can reach, and tests on 600 new shapes it draws itself (Xte is for your own testing). Reach at least 85% within the weight budget.

AI off: no tutor, hints or worked solution on this one. Your own thinking only.

Spotted a problem in this lesson? Report it · Ask about this lesson

This exercise opens soon

Finish the lesson before it and this one unlocks. The whole course is free, no account needed, and you can read this lesson right now.