»ã±¨ÈË£ºRichard¸ß¼¶½²Ê¦[University of Macau, China]
»ã±¨¹¦·ò£º12Ô 01ÈÕ£¨Öܶþ£©9: 00¡«10: 30
»ã±¨µØÖ·£ºÐ£±¾²¿¶«ÇøÍÆËã»úѧԺ´óÂ¥901ÊÒ
Ñû Çë ÈË£ºÂæÏé·å ½ÌÊÚ
Ö÷°ì£ºÍÆËã»úѧԺ
ÄÚÈÝÌáÒª£º
Mixed-Membership Stochastic Block model (MMSB) is a popular framework for modeling social network relationships. However, this model makes an assumption that the distributions of relational membership indicators between the two nodes are independent, which may not be true under many real settings. We introduce a new framework where individual Copula function is to be employed to jointly model the membership pairs of those nodes within the subgroup of interest using Bayesian Non-Parametric methods. We present our model as well as its sampling strategies for both the finite and infinite (number of categories) case.
Brief Biography:
Richard received PhD from University of Technology, Sydney (UTS). He is currently an academic working at the School of Computing and Communications at UTS. He was previously working at School of Computing and Mathematics in Charles Sturt University. He also worked at IBM Australia prior to becoming an academic. Richard has been research active in machine learning, image processing, computer vision and recently statistical data mining. He is currently supervising and co-supervising eight PhD students. He has spent six months during Jan 2012 - Jun 2012 visiting Department of Statistics, Oxford University. In addition to papers, he has spent considerable time writing gentle introduction tutorials on machine learning models.
