Deep Learning · MT2 · guide

The maths behind a network

1. A vector is a list and a matrix is a table
2. Matrix multiplication: every row has a job
3. A graph and a slope
4. Derivatives without a calculus course
5. Derivatives: from an ordinary change to a local slope
6. Partial derivatives hold other things still
7. The chain rule is multiplying linked effects
8. The derivative of prediction error
9. Loss scaling: derive the gradient instead of guessing a factor
10. Probability mean and variance in plain language
11. Mean, variance and standard deviation from ordinary arithmetic
12. Standardization: what the answer 1 actually means
13. Chain rule: the complete calculation