Generalized reduced rank latent factor regression for high dimensional tensor fields, and neuroimaging-genetic applications

详细信息    查看全文

• 作者：Chenyang Tao^a ; ^b ; Thomas E. Nichols^c ; Xue Hua^d ; Christopher R.K. Ching^d ; ^e ; Edmund T. Rolls^b ; ^f ; Paul M. Thompson^d ; ^g ; Jianfeng Feng^a ; ^b ; ^h ; ^{jianfeng64@gmail.com} ; The Alzheimer's Disease Neuroimaging Initiative¹
• 关键词：Dimension reduction ; Generalized linear model ; High dimensional tensor field ; Latent factor ; Least squares kernel machines ; Nuclear norm regularization ; Reduced rank regression ; Riemannian manifold
• 刊名：NeuroImage
• 出版年：2017
• 出版时间：1 January 2017
• 年：2017
• 卷：144
• 期：part_PA
• 页码：35-57
• 全文大小：2648 K
• 卷排序：144

文摘

A new regression model controlling for effective dimensions and latent factors. We show how latent factors jeopardize traditional methods but not this new model. A kernel-based procedure is also proposed for efficient imaging genetic studies. We apply our method to ADNI dataset and CACNA1C to be the most significant gene. Our results show more significance and biological relevance than previous studies.