AD3: alternating directions dual decomposition for MAP inference in graphical models

77 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Structured Convolutional Kernel Networks for Airline Crew Scheduling (2021, methodology): “Unlike for the OCR dataset where “max-product” is used for belief propagation, in this section, we use AD3 (alternating directions dual decomposition) (Martins et al., 2015) for approximate maximum a posteriori (MAP) inference. available and the need for extensive ﬁne-tuning makes a predictor…”
Hinge-Loss Markov Random Fields and Probabilistic Soft Logic (2015, methodology): “This constraint must be active because it is violated by the optimizers of both modified objectives (Martins et al., 2015, Lemma 17).”
SparseMAP: Differentiable Sparse Structured Inference (2018, methodology): “For this reason, we consider the active set method for constrained QPs: a generalization of Wolfe’s min-norm point algorithm (Wolfe, 1976), also used in structured prediction for the quadratic subproblems by Martins et al. (2015).”
Greedy Direction Method of Multiplier for MAP Inference of Large Output Domain (2017, methodology): “Among them the LP-based relaxation exhibits the best tradeoff between accuracy and efficiency (particularly on large-scale problems) and is arguably the most investigated approach in recent years [1, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14].”