Instead of programming rules by hand, feed examples and let the algorithm find patterns. Machine learning builds models that improve with data, and it now underpins search, recommendations, fraud detection, and speech recognition.

The field splits by supervision. Supervised learning uses labeled data. Unsupervised learning finds structure without labels. Reinforcement learning learns from reward signals through interaction. Each suits different problems.

Common learning paradigms

• Supervised learning from labels
• Unsupervised learning from structure
• Reinforcement learning from reward
• Semi-supervised learning from mixed data
• Self-supervised learning from pretext tasks

Data dominates outcomes. A simple model on clean, relevant data often beats a complex model on noisy data. Practitioners spend most of their time on data collection, cleaning, and validation, not on architecture search.