Feature Extraction
In plain terms
Deriving informative inputs from raw data for a model.
Going deeper
Feature extraction transforms raw data into meaningful representations that make learning easier, manually or via learned layers. It surfaces the signal models need.
Under the hood
Classical ML relies on hand-crafted features; deep learning learns them automatically.
Examples
Image features, TF-IDF, embeddings, signal processing, preprocessing.
Think of it like...
Like distilling key clues from a messy crime scene into a useful report.