方法对比
并排查看您选择的方法；存在差异的行会高亮显示。
| 在线线性回归× | 在线学习× | |
|---|---|---|
| 领域 | 机器学习 | 机器学习 |
| 方法族 | Machine learning | Machine learning |
| 起源年份≠ | 1960 (LMS); 1950 (RLS formalization) | 1958–2000s |
| 提出者≠ | Widrow, B. & Hoff, M. E. (LMS); Gauss / Plackett (RLS) | Rosenblatt, F.; Littlestone, N.; Shalev-Shwartz, S. (key contributors) |
| 类型≠ | Incremental supervised regression | Learning paradigm (sequential model update) |
| 开创性文献≠ | Shalev-Shwartz, S. (2012). Online Learning and Online Convex Optimization. Foundations and Trends in Machine Learning, 4(2), 107–194. DOI ↗ | Shalev-Shwartz, S. (2011). Online Learning and Online Convex Optimization. Foundations and Trends in Machine Learning, 4(2), 107–194. DOI ↗ |
| 别名 | incremental linear regression, streaming linear regression, recursive least squares regression, stochastic gradient descent regression | incremental learning, sequential learning, streaming learning, online machine learning |
| 相关 | 6 | 6 |
| 摘要≠ | Online Linear Regression fits a linear model one observation at a time, updating weights incrementally as each new data point arrives. Unlike batch least-squares, it never needs to store or re-process the full dataset, making it the natural choice for streaming data, very large datasets, and environments where the data-generating process can shift over time. | Online learning is a machine learning paradigm in which a model is updated incrementally as each new data point arrives, rather than being trained once on a fixed dataset. It is essential when data streams continuously, storage is limited, or the underlying distribution shifts over time. Theoretical performance is measured by cumulative regret relative to the best fixed predictor in hindsight. |
| ScholarGate数据集 ↗ |
|
|