Name
Abstract | ||
|---|---|---|
Abstract We describe computationally efficient meth- ods for learning mixtures in which each com- ponent is a directed acyclic graphical model (mixtures of DAGs or MDAGs). We argue that simple search-and-score algorithms are infeasible for a variety of problems, and in- troduce a feasible approach in which param- eter and structure search is interleaved and expected data is treated as real data. Our approach can be viewed as a combination of (1) the Cheeseman–Stutz asymptotic ap- proximation for model posterior probability and (2) the Expectation–Maximization algo- rithm. We evaluate our procedure for select- ing among MDAGs on synthetic and real ex- amples. |
Year | Venue | Keywords |
|---|---|---|
2013 | UAI'98 Proceedings of the Fourteenth conference on Uncertainty in artificial intelligence | cheeseman-stutz asymptotic approximation,simple search-and-score algorithm,acyclic graphical model,computationally efficient method,real example,dag model,expected data,feasible approach,expectation-maximization algorithm,model posterior probability,expectation maximization,graphical model,random variable,missing data,posterior probability,model selection,expectation maximization algorithm |
Field | DocType | Volume |
Computer science,Posterior probability,Artificial intelligence,Graphical model,Machine learning | Journal | abs/1301.7415 |
ISBN | Citations | PageRank |
1-55860-555-X | 29 | 32.34 |
References | Authors | |
12 | 4 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Bo Thiesson | 1 | 233 | 79.40 |
| Christopher Meek | 2 | 1770 | 248.06 |
| David Maxwell Chickering | 3 | 2462 | 529.52 |
| David Heckerman | 4 | 6951 | 1419.21 |