Prediction Model of Probability Distribution Based on Observational Learning
LV Zonglei
CHEN Guoming
Abstract：A new prediction model of probability distribution based on observational learning has been proposed in this paper ,which is combined with the concept of loosening control conditions and virtual sample generation .Observational learn‐ing algorithm is expanded to research the probability distribution under small sample in this model ,which is applied to point prediction and classification traditionally .The model extracts the subsets with loosening attribute conditions and creates base learners with cubic spline interpolation function .The virtual samples are used to promote the consistency of base learners e‐ventually .The model provides calculation formula for trust of learner and optimizes the exit mechanism to apply the model better .The results from manual dataset and real world problems from UCI repository shows that the model solves the prob ‐lem of probability distribution prediction under small samples and the optimized observational learning algorithm is better and higher in generalization and precision than before .
Keywords：observational learning algorithmprobability distributionsmall sample size problemvirtual sample gen-eration
Publication Date：2016-01-01
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：7( 1635-1640,1649 )