WEKO3
アイテム
重み付きEM学習とモニタ構造
https://ipsj.ixsq.nii.ac.jp/records/131187
https://ipsj.ixsq.nii.ac.jp/records/131187db0a134a-5882-48ee-97ca-e3537bfc6481
| 名前 / ファイル | ライセンス | アクション |
|---|---|---|
KJ00001344935.pdf (206.0 kB)
|
|
| Item type | National Convention(1) | |||||
|---|---|---|---|---|---|---|
| 公開日 | 1997-03-12 | |||||
| タイトル | ||||||
| タイトル | 重み付きEM学習とモニタ構造 | |||||
| タイトル | ||||||
| 言語 | en | |||||
| タイトル | The Weighted EM Learning and Monitoring Structure | |||||
| 言語 | ||||||
| 言語 | eng | |||||
| 資源タイプ | ||||||
| 資源タイプ識別子 | http://purl.org/coar/resource_type/c_5794 | |||||
| 資源タイプ | conference paper | |||||
| 著者所属 | ||||||
| 早稲田大学理工学部電気電子情報工学科 | ||||||
| 著者所属(英) | ||||||
| en | ||||||
| Department of Electrical, Electronics & Computer Engineering, Waseda University | ||||||
| 論文抄録(英) | ||||||
| 内容記述タイプ | Other | |||||
| 内容記述 | Computing an expectation and maximization is a set of powerful tool in statistical data processing. Dempster et al. [DEM78] collected examples from diverse areas, built a unified theory and coined the name "EM algorithm." Then, Jordan and Jacobs [JOR94] connected their leaning strategy on hierarchical mixtures of experts with this EM algorithm. These algorithms heavily depend on the logarithm and the nonnegativity of the Kullback-Leibler's divergence. Yet, there is a wider clms for such an information measure; the divergence of order α. The paper uses the generalized measure in order to derive a probability weighted EM algorithm and learning strategies. Extended versions of statistics methods such as the Fisher's measure of information and the Cramer-Rao's bound appear in the execution of learning. Finally, usage of the generalized EM algorithm & a building block is discussed. | |||||
| 書誌レコードID | ||||||
| 収録物識別子タイプ | NCID | |||||
| 収録物識別子 | AN00349328 | |||||
| 書誌情報 |
全国大会講演論文集 巻 第54回, 号 人工知能と認知科学, p. 161-162, 発行日 1997-03-12 |
|||||
| 出版者 | ||||||
| 言語 | ja | |||||
| 出版者 | 情報処理学会 | |||||
