Broadcasting is a powerful feature in NumPy (and many deep learning libraries) that lets you perform operations on arrays of different shapes without writing explicit for loops. This makes your code simpler, faster, and more readable.
🧠 What is a Computation Graph?
This notebook explains how to compute derivatives using a computation graph, following the example from Andrew Ng's neural networks course. It breaks down a simple function into smaller steps and shows how to apply the chain rule to calculate how changes in inputs affect the final output.
This page explains how to compute derivatives in logistic regression using a computation graph, broken down into easy-to-understand steps.
🟢 Forward Propagation (Forward Pass)
🌈 The Function: f(x) = x^2
This knowledge item (KI) explains how we move from a loop-based implementation of logistic regression to a clean, fast, vectorized version using NumPy.