Downloads provided by UsageCounts
Dans ce travail, nous étendons une méthode de type Burer-Monteiro pour calculer des minorants convexes (relaxation demi-définie positive) pour le problème de Maximum d'a Posteriori dans des modèles graphiques discrets où les variables ont un nombre d'états et des potentiels binaires arbitraires. Nous considérons une méthode incluant une contrainte pénalisée et une algorithme de type descente en coordonnées par bloc qui évite l’apparition de coefficient de magnitude importante dans la matrice de coûts. Nous montrons que l'algorithme est décroissant. Une évaluation empirique montre que l'approche BCD se compare favorablement à l'approche pénalisée et des minorants linéaires usuels, tels que ceux basés sur des approches de type "message passing" convergentes.
In this paper, we extend a Burer-Monteiro style method to compute low rank Semi-Definite Programming (SDP) bounds for the MAP problem on discrete graphical models with an arbitrary number of states and arbitrary pairwise potentials. We consider both a penalized constraint approach and a dedicated Block Coordinate Descent (BCD) approach which avoids large penalty coefficients in the cost matrix. We show our algorithm is decreasing. Experiments show that the BCD approach compares favorably to the penalized approach and to usual linear bounds relying on convergent message passing approaches.
[INFO.INFO-AI] Computer Science [cs]/Artificial Intelligence [cs.AI], [INFO.INFO-DS] Computer Science [cs]/Data Structures and Algorithms [cs.DS], [MATH.MATH-OC] Mathematics [math]/Optimization and Control [math.OC], [INFO.INFO-LG] Computer Science [cs]/Machine Learning [cs.LG], Convex Optimization, Low rank, Graphical Models, Discrete Optimization, Semi-Definite Programming, Maximum A Posteriori
[INFO.INFO-AI] Computer Science [cs]/Artificial Intelligence [cs.AI], [INFO.INFO-DS] Computer Science [cs]/Data Structures and Algorithms [cs.DS], [MATH.MATH-OC] Mathematics [math]/Optimization and Control [math.OC], [INFO.INFO-LG] Computer Science [cs]/Machine Learning [cs.LG], Convex Optimization, Low rank, Graphical Models, Discrete Optimization, Semi-Definite Programming, Maximum A Posteriori
| selected citations These citations are derived from selected sources. This is an alternative to the "Influence" indicator, which also reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically). | 0 | |
| popularity This indicator reflects the "current" impact/attention (the "hype") of an article in the research community at large, based on the underlying citation network. | Average | |
| influence This indicator reflects the overall/total impact of an article in the research community at large, based on the underlying citation network (diachronically). | Average | |
| impulse This indicator reflects the initial momentum of an article directly after its publication, based on the underlying citation network. | Average |
| downloads | 2 |

Downloads provided by UsageCounts