Catalogue of Artificial Intelligence Techniques
Author(s): R.M. Cameron-Jones
A method, for determining a solution to an ill-posed mathematical problem, by constraining the solution to be, in some sense, regular or smooth. An ill-posed problem is one for which there does not exist a unique stable solution. The best known regularisation method is that of minimising (with respect to the solution sought) a functional which consists of two terms, the first representing the fidelity of the solution to the data, and the second its smoothness using a constraint measure chosen from physical considerations of the original problem.
- Poggio, T., Torre, V. and Koch, C., Computational Vision and Regularisation Theory Nature 317 (1985), 314--319.