Author: Kshirsagar, Meghana; Carbonell, Jaime; Klein-Seetharaman, Judith
Title: Multitask learning for host–pathogen protein interactions Document date: 2013_7_1
ID: sdgt2ms5_70
Snippet: The averaged F1 over the 50 bootstrap experiments for the four tasks with MTPL and Indep. has been tabulated in the Supplementary Table S1. We observe that MTPL does better than Indep. for the three high-throughput datasets and marginally underperforms for the S.typhi. dataset. Table 4 shows the P-values on applying the paired t-tests to the 50 F1 values. For three of the four tasks, the performance improvement by MTPL is clearly statistically si.....
Document: The averaged F1 over the 50 bootstrap experiments for the four tasks with MTPL and Indep. has been tabulated in the Supplementary Table S1. We observe that MTPL does better than Indep. for the three high-throughput datasets and marginally underperforms for the S.typhi. dataset. Table 4 shows the P-values on applying the paired t-tests to the 50 F1 values. For three of the four tasks, the performance improvement by MTPL is clearly statistically significant. For the fourth task, which involves S.typhi., the baseline has a slightly better averaged performance but the P-value does not indicate statistical significance. Hence we can say that the performance of MTPL and Indep. is similar for this task.
Search related documents:
Co phrase search for related documents, hyperlinks ordered by date