Coupling Policy Iteration with Semi-definite Relaxation to Compute Accurate Numerical Invariants in Static Analysis.
Assalé AdjéStéphane GaubertEric GoubaultPublished in: ESOP (2010)
Keyphrases
- static analysis
- semidefinite
- policy iteration
- markov decision processes
- semidefinite programming
- source code
- convex relaxation
- optimal policy
- finite state
- sufficient conditions
- semi definite programming
- fixed point
- regular expressions
- higher dimensional
- model free
- least squares
- reinforcement learning
- infinite horizon
- interior point methods
- convex sets
- temporal difference
- optimal control
- linear programming
- image classification
- machine learning