A Appendix for Details of Deriving HTGM 487 A.1 The lower-bound of the likelihood function

In this section, we provide the details of the lower-bound in Eq. This completes the derivation of Eq. (3). In other words, there is no overlap between any pair of balls. The training algorithm of HTGM is summarized in Algorithm 1. As we discussed in Sec. 2, to the best of our knowledge, our proposed method HTGMis the first B.3 Discussion about the related multi-task learning methods In an MTL method, all tasks are known a priori, i.e., the testing tasks are The second difference lies in the generative process.