Machine learning

Machine learning is a field of artificial intelligence that develops methods for computers to learn patterns from data and use them to make predictions or decisions.

Supervised learning: A machine-learning approach that learns from examples paired with known answers. Labeled examples teach models to predict outcomes such as classes or numerical values.
Empirical risk minimization: A learning principle that selects a model by minimizing its average loss on observed training examples. Many training procedures formalize fitting as minimizing error on available data.
Natural language processing: A field that develops computational methods for analyzing and generating human language. Language models learn patterns from text for translation, classification, and generation.
Rule-based system: A computational system whose behavior is specified through explicit rules written by people. Unlike learned models, its decision logic is directly encoded rather than inferred from examples.
Algorithmic bias: Systematic and unfair differences in the outputs or impacts of computational systems. Training data and modeling choices can reproduce or amplify unequal treatment.
Unsupervised learning: A machine-learning approach that finds structure in data without supplied target labels. It discovers groupings or representations when examples lack known answers.
Loss function: A function that assigns a numerical penalty to a model's prediction errors. Training uses loss to quantify how far predictions are from desired outputs.
Computer vision: A field that develops computational methods for interpreting images and video. Vision systems learn visual patterns for recognition, detection, and measurement.
Statistical inference: The use of data to estimate population properties or evaluate hypotheses under statistical models. Machine learning overlaps with inference but often prioritizes prediction over explanation of parameters.
Explainable artificial intelligence: Methods for making an artificial-intelligence system's outputs or behavior more understandable to people. Complex learned models can be difficult to interpret in consequential settings.

Linked from 47 pages