Timezone: »
EDML is a recently proposed algorithm for learning parameters in Bayesian networks. It was originally derived in terms of approximate inference on a meta-network, which underlies the Bayesian approach to parameter estimation. While this initial derivation helped discover EDML in the first place and provided a concrete context for identifying some of its properties (e.g., in contrast to EM), the formal setting was somewhat tedious in the number of concepts it drew on. In this paper, we propose a greatly simplified perspective on EDML, which casts it as a general approach to continuous optimization. The new perspective has several advantages. First, it makes immediate some results that were non-trivial to prove initially. Second, it facilitates the design of EDML algorithms for new graphical models, leading to a new algorithm for learning parameters in Markov networks. We derive this algorithm in this paper, and show, empirically, that it can sometimes learn better estimates from complete data, several times faster than commonly used optimization methods, such as conjugate gradient and L-BFGS.
Author Information
Khaled Refaat (Waymo)
Arthur Choi (UCLA)
Adnan Darwiche (UCLA)
More from the Same Authors
-
2021 : Accelerated Deep Reinforcement Learning of Terrain-Adaptive Locomotion Skills »
Khaled Refaat · Kai Ding -
2021 : Causal Inference Using Tractable Circuits »
Adnan Darwiche -
2022 : On the Complexity of Counterfactual Reasoning »
Yunqiu Han · Yizuo Chen · Adnan Darwiche -
2017 Poster: Tractability in Structured Probability Spaces »
Arthur Choi · Yujia Shen · Adnan Darwiche -
2016 Poster: Learning Bayesian networks with ancestral constraints »
Eunice Yuh-Jie Chen · Yujia Shen · Arthur Choi · Adnan Darwiche -
2016 Poster: Tractable Operations for Arithmetic Circuits of Probabilistic Models »
Yujia Shen · Arthur Choi · Adnan Darwiche -
2016 Oral: Tractable Operations for Arithmetic Circuits of Probabilistic Models »
Yujia Shen · Arthur Choi · Adnan Darwiche -
2015 Poster: Tractable Learning for Complex Probability Queries »
Jessa Bekker · Jesse Davis · Arthur Choi · Adnan Darwiche · Guy Van den Broeck -
2014 Poster: Decomposing Parameter Estimation Problems »
Khaled Refaat · Arthur Choi · Adnan Darwiche -
2013 Poster: On the Complexity and Approximation of Binary Evidence in Lifted Inference »
Guy Van den Broeck · Adnan Darwiche -
2013 Spotlight: On the Complexity and Approximation of Binary Evidence in Lifted Inference »
Guy Van den Broeck · Adnan Darwiche -
2009 Poster: Approximating MAP by Compensating for Structural Relaxations »
Arthur Choi · Adnan Darwiche