Eisenhut, J., Schuler, X., Fiser, D., Höller, D., Christakis, M., & Hoffmann, J. (2024). New fuzzing biases for action policy testing. In Proceedings of the Thirty-Fourth International Conference on Automated Planning and Scheduling (Issue 1). AAAI Press. http://hdl.handle.net/20.500.12708/200060
Proceedings of the Thirty-Fourth International Conference on Automated Planning and Scheduling
-
ISBN:
978-1-57735-889-3
-
Volume:
34
-
Issue:
1
-
Date (published):
2024
-
Event name:
34th International Conference on Automated Planning and Scheduling (ICAPS 2024)
en
Event date:
1-Jun-2024 - 6-Jun-2024
-
Event place:
Banaff, Alberta, Canada
-
Number of Pages:
6
-
Publisher:
AAAI Press, Washington, DC, USA
-
Peer reviewed:
Yes
-
Keywords:
action policies; policy testing
en
Abstract:
Testing was recently proposed as a method to gain trust in learned action policies in classical planning. Test cases in this setting are states generated by a fuzzing process that performs random walks from the initial state. A fuzzing bias attempts to bias these random walks towards policy bugs, that is, states where the policy performs sub-optimally. Prior work explored a simple fuzzing bias based on policy-trace cost. Here, we investigate this topic more deeply. We introduce three new fuzzing biases based on analyses of policy-trace shape, estimating whether a trace is close to looping back on itself, whether it contains detours, and whether its goal-distance surface does not smoothly decline. Our experiments with two kinds of neural action policies show that these new biases improve bug-finding capabilities in many cases.
en
Project (external):
European Union’s Horizon 2020 German Research Foundation (DFG)