Competitive On-line Linear Regression
نویسنده
چکیده
We apply a general algorithm for merging prediction strategies (the Aggregating Algorithm) to the problem of linear regression with the square loss; our main assumption is that the response variable is bounded. It turns out that for this particular problem the Aggregating Algorithm resembles, but is slightly different from, the wellknown ridge estimation procedure. From general results about the Aggregating Algorithm we deduce a guaranteed bound on the difference between our algorithm's performance and the best, in some sense, linear regression function's performance. We show that the AA attains the optimal constant in our bound, whereas the constant attained by the ridge regression procedure in general can be 4 times worse.
منابع مشابه
Competitive on-line statistics
A radically new approach to statistical modelling, which combines mathematical techniques of Bayesian statistics with the philosophy of the theory of competitive on-line algorithms, has arisen over the last decade in computer science (to a large degree, under the in uence of Dawid's prequential statistics). In this approach, which we call \competitive on-line statistics", it is not assumed that...
متن کاملOptimization of Hydrogen Distribution Network by Imperialist Competitive AlgorithmIn
In modern refineries, hydrogen is widely used for the production of clean fuels. In this paper, a new method is presented in order to use <span style="font-size: 11pt; color: #000000; font-style: norma...
متن کاملThe Role of Psychological Inflexibility and Emotion Dysregulation in Predicting Intensity of Anger in Competitive Athletes
Aim:The purpose of this study was to investigate the role of psychological inflexibility and emotion dysregulation in predicting intensity of anger in competitive athletes. Methods: This was a descriptive and correlational research. The statistical population of this study consisted of all competitive athletes (all athletes who participated in any level of competitive sport, either individual o...
متن کاملFlexible parametric quantile regression model
This article introduces regression quantile models using both RS and FKML generalised lambda distributions (GLD) and demonstrates the versatility of proposed models for a range of linear/non linear and heteroscedastic/homoscedastic empirical data. Owing to the rich shapes of GLDs, GLD quantile regression is a competitive flexible model compared to standard quantile regression. The proposed meth...
متن کاملOrdinal regression neural networks based on concentric hyperspheres
Threshold models are one of the most common approaches for ordinal regression, based on projecting patterns to the real line and dividing this real line in consecutive intervals, one interval for each class. However, finding such one-dimensional projection can be too harsh an imposition for some datasets. This paper proposes a multidimensional latent space representation with the purpose of rel...
متن کامل