\n\n

\nData structured in the form of overlapping or n on-overlapping sets is found in a variety of domains, sometimes explicitly but often subtly. For example, teams, which are of prime importance in indu stry and social science studies are “sets of individuals”; &ldq uo;item sets” in pattern mining of customer transactions are set s and for various types of analysis in language studies a sentence can be c onsidered as a “set or bag of words”. Although building models and inference algorithms for structured data has been an important task in the fields of machine learning and statistics, research on “set-like& rdquo; data still remains less explored. Relationships between pairs of ele ments can be modeled as edges in a graph. However, for modeling relationshi ps that involve all members of a set, hyperedges in a Hypergraph are more n atural representations. This talk describes analyzing complex group structu red data from domains like social networks, customer transaction data and k nowledge graphs, via the lens of Hypergraphs. Specifically:

\n\n Graph v/s Hypergraphs: Motivation for modeling higher-order structures\n Static hypergraphs for modeling higher-order probabilities\n Temporal hypergraphs to capture evolving group structured data with focus on groups in social networks\n Directed hypergraphs for handling directional relationships with applicati ons to Knowledge graphs and Chemistry\n Hyperedge2vec: A representation learning framework for hypergraphs which c an be used for applying deep learning to hypergraphs\n Tool for large-scale distributed computation and predictive modeling for h ypergraph data\n\n DTSTART:20191017T094000 SUMMARY:Hypergraph Analytics: Concepts, Algorithms, and Applications DTEND:20191017T103959 LOCATION: See Description END:VEVENT END:VCALENDAR