Marquardt, D. W. and Snee, R. D. (1975). Ridge regression in practice, American Statistician 29(1): 3-20. Home/Search Document Not in Database Summary Related Articles Check |
This paper is cited in the following contexts:
Bayesian Latent Semantic Analysis - de Freitas, Barnard (2000) (Correct)
....to a single observation. When this happens, the variance typically goes to zero and the likelihood goes to in nity, thus causing serious ill conditioning problems. To circumvent this common problem, people either prune components by hand or add extra tuning parameters as in ridge regression (Marquardt and Snee 1975). From a Bayesian perspective, the introduction of a prior reduces this problem. If, however, the prior is improper, the posterior will also be improper (Robert and Mengersen 1999, Wasserman 1998) That is, there is the possibility that data is not allocated to one of the prior components, so it ....
Marquardt, D. W. and Snee, R. D. (1975). Ridge regression in practice, American Statistician 29(1): 3-20.
Selection of Sufficient Stochastic Data-Models Applied to.. - Moddemeijer, Spaanenburg (1999) (Correct)
....square error by the reduction of the variance dominates the increase by the bias. This means that we can construct better statistics by accepting some bias. In a mean square sense, biased statistics perform even better than unbiased statistics. This concept is closely related to ridge estimation [1 3]. Lets assume two statistics: a conventional statistic b s 1 and a constant statistic b s 2 . The constant statistic is not a statistic in the conventional sense; it s an extreme interpretation of the concept statistic. The constant statistic produces a constant estimate, for example 0, ....
D. W. Marquardt and R. D. Snee, "Ridge regression in practice," The American Statistician, vol. 29, pp. 3--20, 1975.
Selection of Sufficient Stochastic Data-Models Applied to.. - Moddemeijer, Spaanenburg (1999) (Correct)
....square error by the reduction of the variance dominates the increase by the bias. This means that we can construct better statistics by accepting some bias. In a mean square sense, biased statistics perform even better than unbiased statistics. This concept is closely related to ridge estimation [1 3]. Lets assume two statistics: a conventional statistic # s 1 and a constant statistic # s 2 . The constant statistic is not a statistic in the conventional sense; it s an extreme interpretation of the concept statistic. The constant statistic produces a constant estimate, for example 0, ....
D. W. Marquardt and R. D. Snee, "Ridge regression in practice," The American Statistician, vol. 29, pp. 3--20, 1975.
Robust Full Bayesian Learning for Neural Networks - Andrieu, de Freitas, Doucet (1999) (4 citations) (Correct)
....D( 1:k ; x) in which case y 0 1:N;i P i;k y 1:N;i = 0. This event is rather unlikely to occur in our approximation framework, yet we can safeguard against it happening by choosing a very large value for ffi in the simulations. This is a standard least squares trick known as ridge regression (Marquardt and Snee 1975, Wetherill 1986) 5.3 Reversible jump simulated annealing From an MCMC perspective, we can solve the stochastic optimisation problem posed in the previous subsection by adopting a simulated annealing strategy (Geman and Geman 1984, Van Laarhoven and Arts 1987) The simulated annealing method ....
Marquardt, D. W. and Snee, R. D. (1975). Ridge regression in practice, American Statistician 29(1): 3--20.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC