@techreport{oai:ipsj.ixsq.nii.ac.jp:00218747,
 author = {茂木, 亮祐 and 関, 庸一},
 issue = {2},
 month = {Jul},
 note = {混合ガウス分布モデル(Gaussian Mixture Model,GMM)は有限個の多変量正規分布を混合した確率モデルであり,画像認識や音声認識,異常検知,密度推定,クラス分類におけるサブクラスの推定など教師あり・教師なし学習問わず多くの問題で有用なモデルである.GMM の学習における最も困難な課題の一つが混合数の決定であり,様々なアプローチが提案されている.本研究では,クラスタ数推定法の一つである縮小型最尤自己組織化マップ(Shrinking maximum likelihood self-organizing map,SMLSOM)を GMM の学習問題に適用し,数値実験を通じてその有用性を示す., A Gaussian Mixture Model (GMM) is a probabilistic model of a mixture of finite multivariate normal distributions and is helpful for many problems in image recognition, speech recognition, anomaly detection, density estimation, subclass estimation classification, and other unsupervised and supervised learning tasks. One of the most challenging tasks in learning GMMs is determining the number of mixture components, and various approaches have been proposed. In this work, we apply the shrinking maximum likelihood self-organizing map (SMLSOM), one of the methods for estimating the number of clusters, to the GMM learning task and demonstrate its effectiveness through numerical experiments.},
 title = {縮小型最尤自己組織化マップを用いた混合ガウス分布モデルの学習},
 year = {2022}
}