Machine learning is a subset of artificial intelligence focused on building systems that learn patterns from data and improve performance without explicit programming. At its core, the field spans four learning paradigms β supervised, unsupervised, reinforcement, and increasingly self-supervised learning β each addressing distinct problem types. The foundational challenge in ML is the bias-variance tradeoff: simpler models underfit (high bias), complex models overfit (high variance), and the goal is finding the sweet spot where a model generalizes well to unseen data. With the rise of large-scale self-supervised pretraining, understanding when to use fine-tuning versus training from scratch has become equally fundamental.
What This Cheat Sheet Covers
This topic spans 21 focused tables and 152 indexed concepts. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.
Table 1: Learning Paradigms
| Type | Example | Description |
|---|---|---|
X_train, y_train β model.fit(X_train, y_train) | β’ Learns from labeled data where each input has a corresponding output β’ used for classification and regression tasks. | |
model.fit(X) β clusters = model.predict(X) | β’ Finds hidden patterns or structure in unlabeled data β’ common in clustering and dimensionality reduction. | |
X_labeled + X_unlabeled β model.fit(both) | β’ Combines small labeled dataset with large unlabeled data to improve learning β’ bridges supervised and unsupervised approaches. |