Tikhonov regularization
Regularización de Tíjonov
- Date
- 1943-1963 · confidence low
- Place
- Moscow, Russia
- Credited to
- Андрей Николаевич Тихонов
- Field
- Knowledge & symbols
What it is
Tikhonov regularization (or ridge regression) is a mathematical method for estimating regression coefficients when variables are highly correlated, and for solving ill-posed inverse problems by adding a penalty term. Devised by Soviet mathematician Andrey Tikhonov, it is widely used in econometrics, chemistry and engineering.
What it led to
- Ridge regression in statistics and machine learning
- Regularization methods in neural networks (weight decay)
- Stable numerical solution of ill-conditioned inverse problems