Learning in data-limited multimodal scenarios: Scandent decision forests and tree-based features
详细信息 查看全文
- 作者：Soheil Hora ; Mehdi Moradi ; b ; mmoradi@us.ibm.com" class="auth_mail" title="E-mail the corresponding author ; moradi2004@gmail.com" class="auth_mail" title="E-mail the corresponding author
- 关键词：Multimodal data analysis ; Incomplete data analysis ; Decision forest
- 刊名：Medical Image Analysis
- 出版年：2016
- 出版时间：December 2016
- 年：2016
- 卷：34
- 期：Complete
- 页码：30-41
- 全文大小：1009 K
文摘
- •
Introduction of scandent trees as a solution for treating the problem of block wise missing data in multimodal studies.
- •
Introduction of scandent tree-based kernels to transform the features from one modality, based on trees trained on a different modality.
- •
Results on genomics+MRI data for prostate cancer detection and Alzheimer’s disease staging.