Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
162 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Hidden Markov Models and the Bayes Filter in Categorical Probability (2401.14669v2)

Published 26 Jan 2024 in math.ST, cs.SY, eess.SY, math.CT, and stat.TH

Abstract: We use Markov categories to develop generalizations of the theory of Markov chains and hidden Markov models in an abstract setting. This comprises characterizations of hidden Markov models in terms of local and global conditional independences as well as existing algorithms for Bayesian filtering and smoothing applicable in all Markov categories with conditionals. We show that these algorithms specialize to existing ones such as the Kalman filter, forward-backward algorithm, and the Rauch-Tung-Striebel smoother when instantiated in appropriate Markov categories. Under slightly stronger assumptions, we also prove that the sequence of outputs of the Bayes filter is itself a Markov chain with a concrete formula for its transition maps. There are two main features of this categorical framework. The first is its generality, as it can be used in any Markov category with conditionals. In particular, it provides a systematic unified account of hidden Markov models and algorithms for filtering and smoothing in discrete probability, Gaussian probability, measure-theoretic probability, possibilistic nondeterminism and others at the same time. The second feature is the intuitive visual representation of information flow in these algorithms in terms of string diagrams.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (45)
  1. The science of quantitative information flow. Information Security and Cryptography. Springer, Cham, 2020.
  2. Estimation with Applications to Tracking and Navigation: Theory Algorithms and Software. John Wiley & Sons, Inc., 2002. doi:10.1002/0471221279.
  3. G. J. Bierman. Factorization methods for discrete sequential estimation. Mathematics in science and engineering. Academic Press, New York, 1977. ISBN 0120973502.
  4. G. J. Bierman. Square-root information filtering and smoothing for precision orbit determination. In D. C. Sorensen and R. J. B. Wets, editors, Algorithms and Theory in Filtering and Control: Proceedings of the Workshop on Numberical Problems, Part 1, pages 61–75. Springer, Berlin, Heidelberg, 1982. doi:10.1007/BFb0120973.
  5. Inference in Hidden Markov Models. Springer New York, 2005. doi:10.1007/0-387-28982-8.
  6. P. Chigansky and R. van Handel. A Complete Solution to Blackwell’s Unique Ergodicity Problem for Hidden Markov Chains. Ann. Appl. Probab., 20(6):2318–2345, 2010. doi:10.1214/10-AAP688. arXiv:0910.3603.
  7. K. Cho and B. Jacobs. Disintegration and Bayesian inversion via string diagrams. Math. Structures Comput. Sci., 29:938–971, 2019. doi:10.1017/S0960129518000488. arXiv:1709.00322.
  8. B. Fong. Causal Theories: A Categorical Perspective on Bayesian Networks. Master’s thesis, University of Oxford, 2012. arXiv:1301.6201.
  9. T. Fritz. A synthetic approach to Markov kernels, conditional independence and theorems on sufficient statistics. Adv. Math., 370:107239, 2020. doi:10.1016/j.aim.2020.107239. arXiv:1908.07021.
  10. T. Fritz and A. Klingler. The d𝑑ditalic_d-separation criterion in categorical probability. J. Mach. Learn. Res., 24(46):1–49, 2023. url:jmlr.org/papers/v24/22-0916.html. arXiv:2207.05740.
  11. T. Fritz and W. Liang. Free gs-monoidal categories and free Markov categories. Appl. Categ. Structures, 31(21), 2023. doi:10.1007/s10485-023-09717-0. arXiv:2204.02284.
  12. T. Fritz and E. F. Rischel. Infinite products and zero-one laws in categorical probability. Compositionality, 2:3, 2020. doi:10.32408/compositionality-2-3. arXiv:1912.02769.
  13. Absolute continuity, supports and idempotent splitting in categorical probability. arXiv:2308.00651.
  14. Representable Markov categories and comparison of statistical experiments in categorical probability. Theor. Comput. Sci., 961:113896, 2023. doi:10.1016/j.tcs.2023.113896. arXiv:2010.07416.
  15. D. Gao. Functorial causal models: Towards an algebraic approach to causal inference. Bachelor’s thesis, Brown University, 2022. url:brown.edu/academics/math/sites/math/files/Gao%2C%20Dichuan_0.pdf.
  16. B. Gavranović. Space-time tradeoffs of lenses and optics via higher category theory, 2022. arXiv:2209.09351.
  17. Y. Ho and R. Lee. A Bayesian approach to problems in stochastic estimation and control. IEEE Trans. Automat. Contr., 9(4):333–339, 1964. doi:10.1109/TAC.1964.1105763.
  18. B. Jacobs. Structured Probabilistic Reasoning. url:cs.ru.nl/B.Jacobs/PAPERS/ProbabilisticReasoning.pdf.
  19. Causal inference by string diagram surgery. In Int. Conf. Found. Sof. Sci. Comp. Struc., pages 313–329. Springer, 2019. doi:10.1007/978-3-030-17127-8_18.
  20. E. T. Jaynes. Probability theory: The logic of science. Cambridge University Press, 2003. doi:10.1017/CBO9780511790423.
  21. S. Julier and J. Uhlmann. Unscented filtering and nonlinear estimation. Proc. IEEE, 92(3):401–422, 2004. doi:10.1109/JPROC.2003.823141.
  22. New extension of the Kalman filter to nonlinear systems. In Signal processing, sensor fusion, and target recognition VI, volume 3068, pages 182–193. Spie, 1997. doi:10.1117/12.280797.
  23. R. E. Kalman. A new approach to linear filtering and prediction problems. J. Basic Eng., 82:35–45, 1960. doi:10.1115/1.3662552.
  24. New results in linear filtering and prediction theory. J. Basic Eng., 83:95–108, 1961. doi:10.1115/1.3658902.
  25. G. Kitagawa. The two-filter formula for smoothing and an implementation of the gaussian-sum smoother. Annals of the Institute of Statistical Mathematics, 46:605–623, 1994.
  26. J. Lambert. The Bayes filter and intro to state estimation, 2018. url:johnwlambert.github.io/bayes-filter/.
  27. Abstract hidden Markov models: a monadic account of quantitative information flow. Log. Methods Comput. Sci., 15(1):Paper No. 36, 50, 2019.
  28. Unscented Kalman filters for Riemannian state-space systems. IEEE Trans. Automat. Contr., 64(4):1487–1502, 2019. doi:10.1109/TAC.2018.2846684.
  29. Continuous-discrete smoothing of diffusions. Electronic Journal of Statistics, 15, 01 2021. doi:10.1214/21-EJS1894.
  30. K. Murphy. Machine Learning: A Probabilistic Perspective. Adaptive Computation and Machine Learning series. MIT Press, 2012. ISBN 9780262018029. url: dl.acm.org/doi/book/10.5555/2380985.
  31. P. Panangaden. Labelled Markov processes. Imperial College Press, London, 2009. doi:10.1142/p595.
  32. Categorical information flow. In The art of modelling computational systems—a journey from logic and concurrency to security and privacy, volume 11760 of Lecture Notes in Comput. Sci., pages 329–343. Springer, Cham, 2019.
  33. L. Rabiner. A tutorial on hidden Markov models and selected applications in speech recognition. Proc. IEEE, 77(2):257–286, 1989. doi:10.1109/5.18626.
  34. Maximum likelihood estimates of linear dynamic systems. AIAA Journal, 3(8):1445–1450, 1965. doi:10.2514/3.3166.
  35. E. F. Rischel. The category theory of causal models. Master’s thesis, University of Copenhagen, 2020. url:erischel.com/documents/mscthesis.pdf.
  36. E. F. Rischel and S. Weichwald. Compositional abstraction error and a category of causal models. In Proc. Conf. on Uncertainty Artif. Intell., volume 161, pages 1013–1023. MLResearchPress, 2021. arXiv:2103.15758.
  37. Beyond the Kalman Filter: Particle Filters for Tracking Applications. Artech House, 2004. ISBN 9781580536318. url:uk.artechhouse.com/Beyond-the-Kalman-Filter-Particle-Filters-for-Tracking-Applications-P1308.aspx.
  38. S. Roweis and Z. Ghahramani. A unifying review of linear Gaussian models. Neural comput., 11(2):305–345, 1999. doi:10.1162/089976699300016674.
  39. S. Särkkä. Bayesian filtering and smoothing. Cambridge University Press, Cambridge, 2013. doi:10.1017/CBO9781139344203.
  40. M. Schauer and F. van der Meulen. Compositionality in algorithms for smoothing, 2023. arXiv:2303.13865.
  41. E. Seneta. Non-negative matrices and Markov chains. Springer Science & Business Media, 2006. doi:10.1007/0-387-32792-4.
  42. D. Stein. Structural Foundations for Probabilistic Programming Languages. PhD thesis, University of Oxford, 2021. url:dario-stein.de/thesis.pdf.
  43. D. Stein and R. Samuelson. A category for unifying Gaussian probability and nondeterminism. In CALCO, volume 270. LIPIcs, 2023. arXiv:2204.14024.
  44. F. van der Meulen and M. Schauer. Automatic backward filtering forward guiding for markov processes and graphical models, 2022. arXiv:2010.03509.
  45. N. Virgo. Unifilar machines and the adjoint structure of Bayesian models. EPTCS, 397:299–317, 2023. doi:10.4204/EPTCS.397.18. arXiv:2305.02826.

Summary

We haven't generated a summary for this paper yet.