Verifiable RNN-Based Policies for POMDPs Under Temporal Logic Constraints.
Steven CarrNils JansenUfuk TopcuPublished in: IJCAI (2020)
Keyphrases
- temporal logic
- model checking
- partially observable markov decision processes
- automata theoretic
- satisfiability problem
- modal logic
- temporally extended
- dynamic constraints
- optimal policy
- finite state
- concurrent systems
- recurrent neural networks
- nearest neighbor
- reinforcement learning
- computation tree logic
- dynamic programming
- mazurkiewicz traces
- policy search
- linear temporal logic
- belief revision
- markov decision problems
- continuous state
- verification method
- bounded model checking
- partially observable
- transition systems
- formal verification
- belief state
- markov decision processes
- sufficient conditions
- multi agent systems