Online Learning
A learning method where the model updates continuously as new data arrives, one example at a time.
Overview
Online learning is a machine learning approach where the model is trained incrementally as each new data point becomes available, rather than training on the entire dataset at once (batch learning). The model makes a prediction, receives feedback, and immediately updates its parameters.
Key Details
This paradigm is essential for applications with streaming data, such as recommendation systems, fraud detection, and real-time bidding. Algorithms like online gradient descent and the Perceptron are classic examples. The key challenge is balancing adaptability to new patterns with stability of learned knowledge.
Related Concepts
continual learning • stochastic gradient descent • data drift