For deterministic systems, a counterexample to a property can simply be an error trace, whereas counterexamples in probabilistic systems are necessarily more complex. For instance, a set of erroneous traces with a sufficient cumulative probability mass can be used. Since these are too large objects to understand and manipulate, compact representations such as subchains have been considered. In the case of probabilistic systems with non-determinism, the situation is even more complex. While a subchain for a given strategy (or scheduler, resolving non-determinism) is a straightforward choice, we take a different approach. Instead, we focus on the strategy itself, and extract the most important decisions it makes, and present its succinct representation. The key tools we employ to achieve this are (1) introducing a concept of importance of a state w.r.t. the strategy, and (2) learning using decision trees. There are three main consequent advantages of our approach. Firstly, it exploits the quantitative information on states, stressing the more important decisions. Secondly, it leads to a greater variability and degree of freedom in representing the strategies. Thirdly, the representation uses a self-explanatory data structure. In summary, our approach produces more succinct and more explainable strategies, as opposed to e.g. binary decision diagrams. Finally, our experimental results show that we can extract several rules describing the strategy even for very large systems that do not fit in memory, and based on the rules explain the erroneous behaviour.
This research was funded in part by Austrian Science Fund (FWF) Grant No P 23499-N23, FWF NFN Grant No S11407-N23 (RiSE) and Z211-N23 (Wittgenstein Award), European Research Council (ERC) Grant No 279307 (Graph Games), ERC Grant No 267989 (QUAREM), the Czech Science Foundation Grant No P202/12/G061, and People Programme (Marie Curie Actions) of the European Union’s Seventh Framework Programme (FP7/2007–2013) REA Grant No 291734.
158 - 177
CAV: Computer Aided Verification
San Francisco, CA, United States
2015-07-18 – 2015-07-24
Brázdil T, Chatterjee K, Chmelik M, Fellner A, Kretinsky J. Counterexample explanation by learning small strategies in Markov decision processes. In: Vol 9206. Springer; 2015:158-177. doi:10.1007/978-3-319-21690-4_10
Brázdil, T., Chatterjee, K., Chmelik, M., Fellner, A., & Kretinsky, J. (2015). Counterexample explanation by learning small strategies in Markov decision processes (Vol. 9206, pp. 158–177). Presented at the CAV: Computer Aided Verification, San Francisco, CA, United States: Springer. https://doi.org/10.1007/978-3-319-21690-4_10
Brázdil, Tomáš, Krishnendu Chatterjee, Martin Chmelik, Andreas Fellner, and Jan Kretinsky. “Counterexample Explanation by Learning Small Strategies in Markov Decision Processes,” 9206:158–77. Springer, 2015. https://doi.org/10.1007/978-3-319-21690-4_10.
T. Brázdil, K. Chatterjee, M. Chmelik, A. Fellner, and J. Kretinsky, “Counterexample explanation by learning small strategies in Markov decision processes,” presented at the CAV: Computer Aided Verification, San Francisco, CA, United States, 2015, vol. 9206, pp. 158–177.
Brázdil T, Chatterjee K, Chmelik M, Fellner A, Kretinsky J. 2015. Counterexample explanation by learning small strategies in Markov decision processes. CAV: Computer Aided Verification, LNCS, vol. 9206. 158–177.
Brázdil, Tomáš, et al. Counterexample Explanation by Learning Small Strategies in Markov Decision Processes. Vol. 9206, Springer, 2015, pp. 158–77, doi:10.1007/978-3-319-21690-4_10.
Link(s) to Main File(s)
Material in IST: