Selected article for: "current work and important problem"

Author: Kshirsagar, Meghana; Carbonell, Jaime; Klein-Seetharaman, Judith
Title: Multitask learning for host–pathogen protein interactions
  • Document date: 2013_7_1
  • ID: sdgt2ms5_50
    Snippet: Our features define a high-dimensional and sparse space (the number of features is listed in Table 1 ). Because our features are derived by integrating several databases, some of which are not complete, there are many examples and features with missing values. In our current work, we eliminate all examples with 410% missing features. For the rest, we use mean value-based feature imputation. Handling missing data effectively is an important aspect.....
    Document: Our features define a high-dimensional and sparse space (the number of features is listed in Table 1 ). Because our features are derived by integrating several databases, some of which are not complete, there are many examples and features with missing values. In our current work, we eliminate all examples with 410% missing features. For the rest, we use mean value-based feature imputation. Handling missing data effectively is an important aspect of the PPI prediction problem; however, it is not the focus of this work. The remaining examples after elimination and imputation are also shown in Table 1 .

    Search related documents:
    Co phrase search for related documents