Email this article Probabilistic morphisms and Bayesian supervised learning
- Authors: Lê H.V.1,2
-
Affiliations:
- Institute of Mathematics, Czech Academy of Sciences, Praha, Czech Republic
- Faculty of Mathematics and Physics, Charles University, Praha, Czech Republic
- Issue: Vol 216, No 5 (2025)
- Pages: 161-180
- Section: Articles
- URL: https://journals.rcsi.science/0368-8666/article/view/306710
- DOI: https://doi.org/10.4213/sm10191
- ID: 306710
Cite item
Open Access
Access granted
Subscription Access
Abstract
We develop the category theory of Markov kernels to the study of categorical aspects of Bayesian inversions. As a result, we present a unified model for Bayesian supervised learning, including Bayesian density estimation. We illustrate this model with Gaussian process regressions.
About the authors
Hông Vân Lê
Institute of Mathematics, Czech Academy of Sciences, Praha, Czech Republic; Faculty of Mathematics and Physics, Charles University, Praha, Czech Republic
Author for correspondence.
Email: hvle@math.cas.cz
References
- N. Ay, J. Jost, Hông Vân Lê, L. Schwachhöfer, Information geometry, Ergeb. Math. Grenzgeb. (3), 64, Springer, Cham, 2017, xi+407 pp.
- C. M. Bishop, Pattern recognition and machine learning, Inf. Sci. Stat., Springer, New York, 2006, xx+738 pp.
- В. И. Богачев, Гауссовские меры, Наука, М., 1997, 352 с.
- В. И. Богачев, Основы теории меры, т. 1, НИЦ “Регулярная и хаотическая динамика”, М.–Ижевск, 2003, 544 с.
- Н. Н. Ченцов, “Категории математической статистики”, Докл. АН СССР, 164:3 (1965), 511–514
- Н. Н. Ченцов, Статистические решающие правила и оптимальные выводы, Наука, М., 1972, 520 с.
- T. Fritz, “A synthetic approach to Markov kernels, conditional independence and theorems of sufficient statistics”, Adv. Math., 370 (2020), 107239, 105 pp.
- T. Fritz, T. Gonda, P. Perrone, E. F. Rischel, “Representable Markov categories and comparison of statistical experiments in categorical probability”, Theoret. Comput. Sci., 961 (2023), 113896, 37 pp.
- T. Fritz, P. Perrone, S. Rezagholi, “Probability, valuations, hyperspace: three monads on top and the support as a morphism”, Math. Structures Comput. Sci., 31:8 (2021), 850–897
- S. Ghosal, A. van der Vaart, Fundamentals of nonparametric Bayesian inference, Camb. Ser. Stat. Probab. Math., 44, Cambridge Univ. Press, Cambridge, 2017, xxiv+646 pp.
- M. Giry, “A categorical approach to probability theory”, Categorical aspects of topology and analysis (Ottawa, ON, 1980), Lecture Notes in Math., 915, Springer-Verlag, Berlin–New York, 1982, 68–85
- J. Jost, Hông Vân Lê, Tat Dat Tran, “Probabilistic morphisms and Bayesian nonparametrics”, Eur. Phys. J. Plus, 136:4 (2021), 441, 29 pp.
- A. Kock, Commutative monads as a theory of distributions
- P. Koerpernik, F. Pfaff, “Consistency of Gaussian process regression in metric spaces”, J. Mach. Learn. Res., 22 (2021), 244, 27 pp.
- W. F. Lawvere, The category of probabilistic mappings, 1962
- D. Leao, Jr., M. Fragoso, P. Ruffino, “Regular conditional probability, disintegration of probability and Radon spaces”, Proyecciones, 23:1 (2004), 15–29
- N. Morse, R. Sacksteder, “Statistical isomorphism”, Ann. Math. Statist., 37:1 (1966), 203–214
- C. E. Rasmussen, C. K. I. Williams, Gaussian processes for machine learning, Adapt. Comput. Mach. Learn., MIT Press, Cambridge, MA, 2006, xviii+248 pp.
- M. J. Schervish, Theory of statistics, Springer Ser. Statist., Springer-Verlag, New York, 1995, xvi+702 pp.
- V. N. Vapnik, Statistical learning theory, Adapt. Learn. Syst. Signal Process. Commun. Control, John Wiley & Sons, Inc., New York, 1998, xxvi+736 pp.
Supplementary files
