Relative loss bounds for on-line density estimation with the exponential family of distributions

被引:175
作者
Azoury, KS [1 ]
Warmuth, MK
机构
[1] San Francisco State Univ, Coll Business, San Francisco, CA 94132 USA
[2] Univ Calif Santa Cruz, Dept Comp Sci, Santa Cruz, CA 95064 USA
基金
美国国家科学基金会;
关键词
relative loss bounds; worst-case loss bounds; on-line algorithms; exponential family of distributions; linear regression; Bregman divergences;
D O I
10.1023/A:1010896012157
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We consider on-line density estimation with a parameterized density from the exponential family. The on-line algorithm receives one example at a time and maintains a parameter that is essentially an average of the past examples. After receiving an example the algorithm incurs a loss, which is the negative log-likelihood of the example with respect to the current parameter of the algorithm. An off-line algorithm can choose the best parameter based on all the examples. We prove bounds on the additional total loss of the on-line algorithm over the total loss of the best off-line parameter. These relative loss bounds hold for an arbitrary sequence of examples. The goal is to design algorithms with the best possible relative loss bounds. We use a Bregman divergence to derive and analyze each algorithm. These divergences are relative entropies between two exponential distributions. We also use our methods to prove relative loss bounds for linear regression.
引用
收藏
页码:211 / 246
页数:36
相关论文
共 45 条
[1]  
Amari S., 1985, DIFFERENTIAL GEOMETR
[2]  
AUER P, 1995, P 1995 NEUR INF PROC, P316
[3]  
Azoury KS, 1999, UNCERTAINTY IN ARTIFICIAL INTELLIGENCE, PROCEEDINGS, P31
[4]  
BARNDORFFNIELSE.O, 1978, INFORMATION EXPONENT
[5]  
Beckenbach E. F., 1971, INEQUALITIES
[6]  
Blackwell D., 1956, PAC J MATH, V6, P1, DOI [10.2140/pjm.1956.6.1, DOI 10.2140/PJM.1956.6.1]
[7]  
Bregman LM, 1967, USSR Computational Mathematics and Mathematical Physics, V7, P200
[8]   AN ITERATIVE ROW-ACTION METHOD FOR INTERVAL CONVEX-PROGRAMMING [J].
CENSOR, Y ;
LENT, A .
JOURNAL OF OPTIMIZATION THEORY AND APPLICATIONS, 1981, 34 (03) :321-353
[9]   How to use expert advice [J].
CesaBianchi, N ;
Freund, Y ;
Haussler, D ;
Helmbold, DP ;
Schapire, RE ;
Warmuth, MK .
JOURNAL OF THE ACM, 1997, 44 (03) :427-485
[10]   Worst-case quadratic loss bounds for prediction using linear functions and gradient descent [J].
CesaBianchi, N ;
Long, PM ;
Warmuth, MK .
IEEE TRANSACTIONS ON NEURAL NETWORKS, 1996, 7 (03) :604-619