Track 4 Session 3 Spotlights

00:05 Sparse Logistic Regression Learns All Discrete Pairwise Graphical Models - Shanshan Wu 04:13 Fast Convergence of Belief Propagation to Global Optima: Beyond Correlation Decay - Frederic Koehler 08:50 Smoothing Structured Decomposable Circuits - Andy Shih 13:18 Counting the Optimal Solutions in Graphical Models - Radu Marinescu 17:45 Combining Generative and Discriminative Models for Hybrid Inference - Victor Garcia Satorras