Talks, Presentations and Seminars

Abstract: Evaluation of deep reinforcement learning (RL) is inherently challenging. Especially the opaqueness of learned policies and the stochastic nature of both agents and environments make testing the behavior of deep RL agents difficult. We present a search-based testing framework that enables a wide range of novel analysis capabilities for evaluating the safety and performance of deep RL agents. For safety testing, our framework utilizes a search algorithm that searches for a reference trace that solves the RL task. The backtracking states of the search, called boundary states, pose safety-critical situations. We create safety test-suites that evaluate how well the RL agent escapes safety-critical situations near these boundary states. For robust performance testing, we create a diverse set of traces via fuzz testing. These fuzz traces are used to bring the agent into a wide variety of potentially unknown states from which the average performance of the agent is compared to the average performance of the fuzz traces. We apply our search-based testing approach on RL for Nintendo's Super Mario Bros.

BibTex:

@inproceedings{ijcai2022p0072,
  title     = {Search-Based Testing of Reinforcement Learning},
  author    = {Tappler, Martin and Cano Córdoba, Filip and Aichernig, Bernhard K. and Könighofer, Bettina},
  booktitle = {Proceedings of the Thirty-First International Joint Conference on
               Artificial Intelligence, {IJCAI-22}},
  publisher = {International Joint Conferences on Artificial Intelligence Organization},
  editor    = {Lud De Raedt},
  pages     = {503--510},
  year      = {2022},
  month     = {7},
  note      = {Main Track},
  doi       = {10.24963/ijcai.2022/72},
  url       = {https://doi.org/10.24963/ijcai.2022/72},
}

Abstract: A classic introduction to polytope theory is presented, serving as the foundation to develop more advanced theoretical tools, namely the algebra of polyhedra and the use of valuations. The main theoretical objective is the construction of the so called Berline-Vergne valuation. Most of the theoretical development is aimed towards this goal. A little survey on Ehrhart positivity is presented, as well as some calculations that lead to conjecture that generalized permutohedra have positive coefficients in their Ehrhart polynomials. Throughout the thesis three different proofs of Ehrhart's theorem are presented, as an application of the new techniques developed.

BibTex:

@mastersthesis{cano2019introduction,
  title={An Introduction to Polytope Theory through Ehrhart's Theorem},
  author={Cano C{\'o}rdoba, Filip},
  type={M.S. thesis},
  year={2019},
  school={Universitat Polit{\`e}cnica de Catalunya}
}