This program implements a learning-based framework for reachability learning having two modules:
- sampling pair of nodes with most reachability information
- reachability learning.
Our main contribution is to show that our framework is efficient, adaptive and flexible to use different methods to sample the pair of nodes with most reachability information; i.e., the first module. To this end, we propose an approximate version of representative reachability processing algorithms in each category. Reachability_learning.py is the automated code to run different reachability calculation algorithms.