Merge-and-Shrink Abstraction: A Method for Generating Lower Bounds in Factored State Spaces.
Malte HelmertPatrik HaslumJörg HoffmannRaz NissimPublished in: J. ACM (2014)
Keyphrases
- lower bound
- objective function
- high accuracy
- cost function
- theoretical analysis
- state space
- experimental evaluation
- synthetic data
- preprocessing
- clustering method
- computational cost
- dynamic programming
- segmentation method
- high precision
- search methods
- data sets
- generation method
- detection method
- model selection
- significant improvement
- search algorithm
- similarity measure
- machine learning