TY - GEN

T1 - Universal sequential learning and decision from individual data sequences

AU - Merhav, Neri

AU - Feder, Meir

PY - 1992

Y1 - 1992

N2 - Sequential learning and decision algorithms are investigated, with various application areas, under a family of additive loss functions for individual data sequences. Simple universal sequential schemes are known, under certain conditions, to approach optimality uniformly as fast as n-1 log n, where n is the sample size. For the case of finite-alphabet observations, the class of schemes that can be implemented by finite-state machines (FSM's) is studied. It is shown that Markovian machines with sufficiently long memory exist that are asymptotically nearly as good as any given FSM (deterministic or randomized) for the purpose of sequential decision. For the continuous-valued observation case, a useful class of parametric schemes is discussed with special attention to the recursive least squares (RLS) algorithm.

AB - Sequential learning and decision algorithms are investigated, with various application areas, under a family of additive loss functions for individual data sequences. Simple universal sequential schemes are known, under certain conditions, to approach optimality uniformly as fast as n-1 log n, where n is the sample size. For the case of finite-alphabet observations, the class of schemes that can be implemented by finite-state machines (FSM's) is studied. It is shown that Markovian machines with sufficiently long memory exist that are asymptotically nearly as good as any given FSM (deterministic or randomized) for the purpose of sequential decision. For the continuous-valued observation case, a useful class of parametric schemes is discussed with special attention to the recursive least squares (RLS) algorithm.

UR - http://www.scopus.com/inward/record.url?scp=0026978928&partnerID=8YFLogxK

U2 - 10.1145/130385.130430

DO - 10.1145/130385.130430

M3 - ???researchoutput.researchoutputtypes.contributiontobookanthology.conference???

AN - SCOPUS:0026978928

SN - 089791497X

SN - 9780897914970

T3 - Proceedings of the Fifth Annual ACM Workshop on Computational Learning Theory

SP - 413

EP - 427

BT - Proceedings of the Fifth Annual ACM Workshop on Computational Learning Theory

PB - Association for Computing Machinery (ACM)

Y2 - 27 July 1992 through 29 July 1992

ER -