TY - EJOU
AU - Ni, He
AU - Wang, Yongqiao
AU - Xu, Buyun
TI - Self-Organizing Gaussian Mixture Map Based on Adaptive Recursive Bayesian Estimation
T2 - Intelligent Automation \& Soft Computing
PY - 2020
VL - 26
IS - 2
SN - 2326-005X
AB - The paper presents a probabilistic clustering approach based on self-organizing
learning algorithm and recursive Bayesian estimation. The model is built upon
the principle that the market data space is multimodal and can be described by
a mixture of Gaussian distributions. The model parameters are approximated by
a stochastic recursive Bayesian learning: searches for the maximum a posterior
solution at each step, stochastically updates model parameters using a “dualneighbourhood” function with adaptive simulated annealing, and applies profile
likelihood confidence interval to avoid prolonged learning. The proposed model
is based on a number of pioneer works, such as Mixture Gaussian
Autoregressive Model, Self-Organizing Mixture Map, and have some favoured
attributes on its robust convergence and good generalization. The experimental
results on both artificial and real market data show that the algorithm is a good
alternative in measuring multimodal distribution.
KW - Adaptive simulated annealing
KW - Density approximation
KW - Mixture distribution
KW - Profile likelihood confidence interval
KW - Self-Organizing Map
DO - 10.31209/2019.100000068