Skip to main content

Table 4 Performance of FCD features

From: Learning an enriched representation from unlabeled data for protein-protein interaction extraction

ID Local SP-EDF CP-EDF DS-EDF Linear RBF Precision Recall F-score AUC
1 *     *   62.13 62.40 61.4 86.11
2   *    *   53 64.48 57.6 81.75
3   *     * 58.08 62.67 59.16 82.97
4    *   *   56.83 53.55 54.33 79.26
5    *    * 54.64 57.71 54.47 78.34
6     * *   53.5 56.4 53.87 78.02
7     *   * 54.45 58.71 55.56 79.79
8   * * * *   54.08 63.66 58.06 82.61
9   * * * * * 59.7 61.52 60.06 83.78
10 * * * * * * 60.47 68.31 63.54 87.24
  1. In RBF SVM, SVD is used to reduce the feature dimension and a SVM with RBF kernel is used to classify examples. Run 10 combines the outputs of Run 1 and Run 9.