A novel solver for approximate marginal map inference

There is a deep connection between planning and inference, and over the last decade, multiple researchers have introduced explicit reductions showing how stochastic planning can be solved using probabilistic inference with applications in robotics, scheduling, and environmental problems. However, heuristic methods and search are still the best-performing approaches for planning in large combinatorial state and action spaces. My co-authors and I take a new approach in our paper, "From Stochastic Planning to Marginal MAP" (authors: Hao Cui, Radu Marinescu, Roni Khardon), at the 2018 Conference on Neural Information Processing Systems (NeurIPS) by showing how ideas from planning can be used for inference.

Read the whole article on Physorg

More from Physorg