Cross-disciplinary research between machine learning and biostatistics based on curve estimation

基于曲线估计的机器学习与生物统计学的跨学科研究

• 批准号: 23500350
• 负责人: NAITO Kanta
• 金额: $ 3.33万
• 依托单位: Shimane University
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (C)
• 财政年份: 2011
• 资助国家: 日本
• 起止时间: 2011 至 2013
• 项目状态: 已结题
• 来源: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-23500350/
• 关键词: 平滑化; 関数推定; パターン認識; 機械学習; 高次元小標本; 生物統計; 漸近理論; 特徴選択; 高次元

As a cross-displinary research between machine learning and biostatistics, I have been addressing Deepened Research (Theoretical Research), Expanded Research(Developing Methodology), and Applied Research, with referring to the given effort. Five papers have been published in Deepened Research, which is the progress more than expected. I have published three papers in Expanded Research, so it certainly got the progress.In Applied Research, asymptotic distribution of dilatation of a certain quasi-conformal mapping was developed, which can be applied toanalysis of human fetus data. The LMS method, which is one of efficent nonlinear regression methods, has been extended tononlinear multivariate regression setting. The resultant method is called the LMSR method, and it has been applied to analysis of human fetus data.

作为机器学习与生物统计学之间的跨分离研究，我一直在处理深度研究（理论研究），扩展研究（开发方法）和应用研究，指的是指定的努力。五篇论文已经发表在深入的研究中，这比预期的要多。我已经在扩展研究中发表了三篇论文，因此它肯定取得了进步。在应用研究中，开发了某个准统一形式映射的扩张的渐近分布，这可以对人类胎儿数据进行纳分析。 LMS方法是有效的非线性回归方法之一，已被扩展为Tononlinear多元回归设置。所得方法称为LMSR方法，已应用于人类胎儿数据的分析。

项目成果

Statistical Analysis with Dilatation for development Process of Human Fetuses

人类胎儿发育过程的扩张统计分析

• 发表时间: 2011
• 作者: 内藤貫太;野津昭文;宇田川潤;大谷浩
• 通讯作者: 大谷浩

高次元小標本におけるnaïve canonical correlation に基づく多群判別-特徴選択

高维小样本中基于朴素典型相关性的多组判别-特征选择

• 发表时间: 2013
• 作者: 内藤貫太;玉谷充;Inge Koch
• 通讯作者: Inge Koch

Multi-class discriminant function based on canonical correlation in high dimension low sample size

高维低样本下基于典型相关的多类判别函数

• 发表时间: 2013
• 期刊: Bulletin of Informatics and Cebernetics
• 作者: Mitsuru Tamatani;Kanta Naito and Inge Koch
• 通讯作者: Kanta Naito and Inge Koch

Selection of smoothing parameter for one-step sparse estimates with L_q penalty

具有 L_q 罚分的一步稀疏估计的平滑参数选择

• 发表时间: 2011
• 期刊: Journal of Data Science
• 作者: Kim;S.;Hayashi;K. and Kurihara;K.;Masaru Kanba and Kanta Naito
• 通讯作者: Masaru Kanba and Kanta Naito

Density estimation with minimization of U-divergence

最小化 U 发散的密度估计

• 发表时间: 2012
• 作者: Ogasawara;H;Kanta Naito and Shinto Eguchi
• 通讯作者: Kanta Naito and Shinto Eguchi