Abstract |
In the study of bioinformatics, one important problem is the prediction of clinical outcomes using profiling datasets with a large amount of variables such as gene expression data, proteomics data and metabolomics data. In such datasets, major challenges lie in the relatively small number of samples compared to the large number of predictors namely the 'n «p' issue. In this talk, the conceptualization of Geometric Deep Learning will be presented which provides a firm ground to deal with 'n <<< p' issue.
|