Online Gaussian Mixture Model for Concept Modeling and Discovery

Chun-Sheng Fang ； Anca L. Ralescu

《International Journal of Intelligent Technologies and Applied Statistics》 2卷1期 (2009/03) Pp. 59-75

https://doi.org/10.6148/IJITAS.2009.0201.05

並列摘要

Concept discovery and modeling are fundamental problems in machine learning research. Real world concepts are usually high-dimensional and have complicated distributions along their dimensions. Gaussian Mixture Models (GMM) have proved useful in modeling such complicated distributions. We propose a data-driven concept modeling and discovery framework using GMM, with on-line updating mechanism for fast computation suitable for real world applications. Experiments show the efficacy and efficiency of the proposed algorithm.

並列關鍵字

Conceptual spaces ； Gaussian mixed models