# Inferring finite automata with stochastic output functions and an application to map learning

- 331 Downloads
- 11 Citations

## Abstract

It is often useful for a robot to construct a spatial representation of its environment from experiments and observations, in other words, to learn a map of its environment by exploration. In addition, robots, like people, make occasional errors in perceiving the spatial features of their environments. We formulate map learning as the problem of inferring from noisy observations the structure of a reduced deterministic finite automaton. We assume that the automaton to be learned has a distinguishing sequence. Observation noise is modeled by treating the observed output at each state as a random variable, where each visit to the state is an independent trial and the correct output is observed with probability exceeding 1/2. We assume no errors in the state transition function.

Using this framework, we provide an exploration algorithm to learn the correct structure of such an automaton with probability 1−δ, given as inputs δ, an upper bound*m* on the number of states, a distinguishing sequence*s*, and a lower bound α>1/2 on the probability of observing the correct output at any state. The running time and the number of basic actions executed by the learning algorithm are bounded by a polynomial in δ^{−1},*m*, |*s*|, and (1/2−α)^{−1}.

We discuss the assumption that a distinguishing sequence is given, and present a method of using a weaker assumption. We also present and discuss simulation results for the algorithm learning several automata derived from office environments.

## Keywords

Automata inference noisy outputs distinguishing sequences map learning spatial representation## References

- Angluin, D. (1978). On the complexity of minimum inference of regular sets.
*Information and Control*39:337–350.Google Scholar - Angluin, D. (1987). Learning regular sets from queries and counterexamples.
*Information and Computation*75:87–106.Google Scholar - Aslam, J.A., & Rivest, R.L. (1990). Inferring graphs from walks. In
*Proceedings COLT-88*.Google Scholar - Bachrach, J.R. (1992). Connectionist modeling and control of finite state environments. Technical Report 92-6, University of Massachusetts at Amherst Department of Computer and Information Science.Google Scholar
- Basye, K., & Dean, T. (1989). Map learning with indistinguishable locations. In
*Proceedings of the Fifth Workshop on Uncertainty in AI*. 7–13.Google Scholar - Basye, K., Dean, T., & Vitter, J.S. (1989). Coping with uncertainty in map learning. In
*Proceedings IJCAI 11*. IJCAII, 663–668.Google Scholar - Basye, K., Dean, T., & Kaelbling, L. (to appear). Learning dynamics: System identification for perceptually challenged agents.
*Artificial Intelligence*.Google Scholar - Basye, K. (1992). A framework for map construction. Technical report, Brown University Department of Computer Science, Providence, RI.Google Scholar
- Dean, T., Basye, K., Chekaluk, R., Hyun, S., Lejter, M., & Randazza, M. (1990). Coping with uncertainty in a control system for navigation and exploration. In
*Proceedings AAAI-90*. AAAI. 1010–1015.Google Scholar - Donald, B., & Jennings, J. (1991). Sensor interpretation and task-directed planning using perceptual equivalence classes. Technical Report CU-CS-TR, Cornell University Computer Science Department.Google Scholar
- Dudek, G., Jenkins, M., Milios, E., & Wilkes, D. (1988). Robotic exploration as graph construction. Technical Report RBCV-TR-88-23, University of Toronto.Google Scholar
- Gold, E.M. (1972). System identification via state characterization.
*Automatica*8:621–636.Google Scholar - Gold, E.M. (1978). Complexity of automation identification from given sets.
*Information and Control*37:302–320.Google Scholar - Goodwin, G.C., & Sin, K.S. (1984).
*Adaptive Filtering Prediction and Control*. Prentice-Hall, Englewood Cliffs, New Jersey.Google Scholar - Kaelbling, L., Basye, K., Dean, T., Kokkevis, E., & Maron, O. (1992). Robot map-learning as learning labeled graphs from noisy data. Technical Report CS-92-15, Brown University Department of Computer Science.Google Scholar
- Kearns, M., & Valiant, L.G. (1989). Cryptographic limitations on learning boolean functions and finite automata. In
*Proceedings of the Twenty First Annual ACM Symposium on Theoretical Computing*. 433–444.Google Scholar - Kuipers, B.J., & Byun, Y.-T. (1988). A robust, qualitative method for robot spatial reasoning. In
*Proceedings AAAI-88*. AAAI. 774–779.Google Scholar - Kuipers, B. (1978). Modeling spatial knowledge.
*Cognitive Science*2:129–153.Google Scholar - Levinson, S.E., Rabiner, L.R., & Sondhi, M.M. (1983). An introduction to the application of the theory of probabilistic functions of a Markov process to automatic speech recognition.
*The Bell System Technical Journal*62(4):1035–1074.Google Scholar - Levitt, T.S., Lawton, D.T., Chelberg, D.M., & Nelson, P.C. (1987). Qualitative landmark-based path planning and following. In
*Proceedings AAAI-87*. AAAI. 689–694.Google Scholar - Moore, E.F. (1956). Gedanken-experiments on sequential machines. In
*Automata Studies*. Princeton University Press, Princeton, New Jersey. 129–153.Google Scholar - Pitt, L., & Warmuth, M.K. (1989). The minimum consistent DFA problem cannot be approximated within any polynomial. In
*Proceedings of the Twenty First Annual ACM Symposium on Theoretical Computing*. 421–432.Google Scholar - Rimey, R.D., & Brown, C.M. (1991). Controlling eye movements with hidden Markov models.
*International Journal of Computer Vision*7(1):47–65.Google Scholar - Rivest, R.L., & Schapire, R.E. (1987). Diversity-based inference of finite automata. In
*Proceedings of the Twenty Eighth Annual Symposium on Foundations of Computer Science*. 78–87.Google Scholar - Rivest, R.L., & Schapire, R.E. (1989). Inference of finite automata using homing sequences. In
*Proceedings of the Twenty First Annual ACM Symposium on Theoretical Computing*. 411–420.Google Scholar - Rudich, S. (1985). Inferring the structure of a Markov chain from its output. In
*Proceedings of the Twenty Sixth Annual Symposium on Foundations of Computer Science*. 321–326.Google Scholar - Schapire, R.E. (1991). The design and analysis of efficient learning algorithms. Technical Report MIT/LCS/TR-493, MIT Laboratory for Computer Science.Google Scholar
- Tzeng, W.-G. (1992). Learning probabilistic automata and Markov chains via queries.
*Machine Learning*8:151–166.Google Scholar - Valiant, L.G. (1984). A theory of the learnable.
*Communications of the ACM*27:1134–1142.Google Scholar - Whitehead, S.D., & Ballard, D.H. (1991). Learning to perceive and act by trial and error.
*Machine Learning 7*.Google Scholar - Widrow, B., & Stearns, S.D. (1985).
*Adaptive Signal Processing*. Prentice-Hall, Englewood Cliffs, N.J.Google Scholar - Yannakakis, M., & Lee, D. (1991). Testing finite state machines. In
*ACM Symposium on Theoretical Computing*. ACM Press. 476–485.Google Scholar