Coupling policy iteration with semi-definite relaxation to compute accurate numerical invariants in static analysis
Assalé AdjéStéphane GaubertEric GoubaultPublished in: Log. Methods Comput. Sci. (2012)
Keyphrases
- static analysis
- semidefinite
- policy iteration
- markov decision processes
- semidefinite programming
- convex relaxation
- source code
- model free
- fixed point
- reinforcement learning
- linear programming
- sufficient conditions
- semi definite programming
- higher dimensional
- optimal policy
- least squares
- interior point methods
- regular expressions
- finite state
- test suite
- dynamic programming
- convex sets
- learning algorithm
- markov decision process
- test cases
- semi supervised
- training data