Backpropagation

What it is

Backpropagation is the algorithm used to train neural networks. It calculates how much each weight contributed to the error in the output, then adjusts weights to reduce that error. The "back" refers to working backwards from the output through all layers to determine which weights to change and by how much.

Why it matters

Backpropagation is why neural networks can learn. Without it, there would be no way to systematically improve model performance. It's a foundational algorithm you'll see referenced in AI discussions—understanding it means understanding how models actually learn from examples.