Iterated logarithmic expansions of the pathwise code lengths for exponential families
نویسندگان
چکیده
Rissanen's Minimum Description Length (MDL) principle is a statistical modeling principle motivated by coding theory. For exponential families we obtain pathwise expansions, to the constant order, of the predictive and mixture code lengths used in MDL. The results are useful for understanding diierent MDL forms.
منابع مشابه
کارآیی هشت مدل ریاضی در توصیف اندازه ذرات در برخی خاکهای استان چهارمحال و بختیاری
Selecting an appropriate particle size distribution (PSD) model for a particular soil may be important for a precise estimation of soil hydraulic properties. Various models have been proposed for describing soil PSDs. The objective of this study was to compare the quality of fitting of eight PSD models (Fredlund, Gompertz, van Genuchten, Jaki, Logarithmic, Exponential, Logarithmic-Exponential a...
متن کاملStochastic functional population dynamics with jumps
In this paper we use a class of stochastic functional Kolmogorov-type model with jumps to describe the evolutions of population dynamics. By constructing a special Lyapunov function, we show that the stochastic functional differential equation associated with our model admits a unique global solution in the positive orthant, and, by the exponential martingale inequality with jumps, we dis...
متن کاملSome Probability Inequalities for Quadratic Forms of Negatively Dependent Subgaussian Random Variables
In this paper, we obtain the upper exponential bounds for the tail probabilities of the quadratic forms for negatively dependent subgaussian random variables. In particular the law of iterated logarithm for quadratic forms of independent subgaussian random variables is generalized to the case of negatively dependent subgaussian random variables.
متن کاملPreface: 'Recent advances in the numerical approximation of stochastic partial differential equations'
The numerical solution of stochastic partial differential equations (SPDEs) is at a stage of development roughly similar to that of stochastic ordinary differential equations (SODEs) in the 1970s, when stochastic Taylor schemes based on an iterated application of the Itô formula were introduced and used to derive higher order numerical schemes. An Itô formula in the generality needed for Taylor...
متن کاملStatistical Curvature and Stochastic Complexity
We discuss the relationship between the statistical embedding curvature [1, 2] and the logarithmic regret [11] (regret for short) of the Bayesian prediction strategy (or coding strategy) for curved exponential families and Markov models. The regret of a strategy is defined as the difference of the logarithmic loss (code length) incurred by the strategy and that of the best strategy for each dat...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید
ثبت ناماگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید
ورودعنوان ژورنال:
- IEEE Trans. Information Theory
دوره 46 شماره
صفحات -
تاریخ انتشار 2000