Name
Abstract | ||
|---|---|---|
Event datasets involve irregular occurrences of events over the timeline and are prevalent in numerous domains. We introduce proximal graphical event models (PGEMs) as a representation of such datasets. PGEMs belong to a broader family of graphical models that characterize relationships between various types of events; in a PGEM, the rate of occurrence of an event type depends only on whether or not its parents have occurred in the most recent history. The main advantage over state-of-the-art models is that learning is entirely data driven and without the need for additional inputs from the user, which can require knowledge of the domain such as choice of basis functions and hyper-parameters. We theoretically justify our learning of parental sets and their optimal windows, proposing sound and complete algorithms in terms of parent structure learning. We present efficient heuristics for learning PGEMs from data, demonstrating their effectiveness on synthetic and real datasets. |
Year | Venue | Keywords |
|---|---|---|
2018 | ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 31 (NIPS 2018) | constant factors,the state of the art,basis functions,recent history,belong to,data driven,proximal graphical event models |
Field | DocType | Volume |
Data-driven,Event type,Hyperparameter,Computer science,Timeline,Heuristics,Artificial intelligence,Basis function,Parent structure,Machine learning | Conference | 31 |
ISSN | Citations | PageRank |
1049-5258 | 0 | 0.34 |
References | Authors | |
0 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Debarun Bhattacharjya | 1 | 59 | 14.91 |
| Dharmashankar Subramanian | 2 | 28 | 8.22 |
| Tian Gao | 3 | 1 | 4.07 |