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

Model orthogonalization and Bayesian forecast mixing via Principal Component Analysis (2405.10839v2)

Published 17 May 2024 in nucl-th, physics.data-an, and stat.ML

Abstract: One can improve predictability in the unknown domain by combining forecasts of imperfect complex computational models using a Bayesian statistical machine learning framework. In many cases, however, the models used in the mixing process are similar. In addition to contaminating the model space, the existence of such similar, or even redundant, models during the multimodeling process can result in misinterpretation of results and deterioration of predictive performance. In this work we describe a method based on the Principal Component Analysis that eliminates model redundancy. We show that by adding model orthogonalization to the proposed Bayesian Model Combination framework, one can arrive at better prediction accuracy and reach excellent uncertainty quantification performance.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (53)
  1. The hydrologist’s guide to Bayesian model selection, averaging and combination. \JournalTitleJ. Hydrol. 572, 96–107, DOI: https://doi.org/10.1016/j.jhydrol.2019.01.072 (2019).
  2. Using Stacking to Average Bayesian Predictive Distributions (with Discussion). \JournalTitleBayesian Anal. 13, 917 – 1007, DOI: https://doi.org/10.1214/17-BA1091 (2018).
  3. Phillips, D. R. et al. Get on the BAND wagon: A Bayesian framework for quantifying model uncertainties in nuclear dynamics. \JournalTitleJ. Phys. G 48, 072001, DOI: https://doi.org/10.1088/1361-6471/abf1df (2021).
  4. Bayesian model averaging: A tutorial. \JournalTitleStatist. Sci. 14, 382–401, DOI: https://doi.org/10.1214/ss/1009212519 (1999).
  5. Wasserman, L. Bayesian model selection and model averaging. \JournalTitleJ. Math. Psych. 44, 92–107, DOI: https://doi.org/10.1006/jmps.1999.1278 (2000).
  6. Bayesian model averaging: A systematic review and conceptual classification. \JournalTitleInt. Stat. Rev. 86, 1–28, DOI: https://doi.org/10.1111/insr.12243 (2018).
  7. A Bayes Interpretation of Stacking for ℳℳ\mathcal{M}caligraphic_M-Complete and ℳℳ\mathcal{M}caligraphic_M-Open Settings. \JournalTitleBayesian Anal 12, 807 – 829, DOI: https://doi.org/10.1214/16-BA1023 (2017).
  8. Bayesian Hierarchical Stacking: Some Models Are (Somewhere) Useful. \JournalTitleBayesian Anal. 17, 1043 – 1071, DOI: https://doi.org/10.1214/21-BA1287 (2022).
  9. Modelling Spatially Correlated Data via Mixtures: A Bayesian Approach. \JournalTitleJ. R. Stat. Soc., B: Stat. Methodol. 64, 805–826, DOI: https://doi.org/10.1111/1467-9868.00362 (2002).
  10. Local Bayesian Dirichlet mixing of imperfect models. \JournalTitleSci. Rep. 13, 19600, DOI: https://doi.org/10.1038/s41598-023-46568-0 (2023).
  11. The combination of forecasts. \JournalTitleJ. Oper. Res. Soc. 20, 451–468, DOI: https://doi.org/10.1057/jors.1969.103 (1969).
  12. Minka, T. P. Bayesian model averaging is not model combination (2002). Urlhttps://api.semanticscholar.org/CorpusID:116598428.
  13. Improved methods of combining forecasts. \JournalTitleJ. Forecast. 3, 197–204, DOI: https://doi.org/10.1002/for.3980030207 (1984).
  14. Turning Bayesian model averaging into Bayesian model combination. In The 2011 International Joint Conference on Neural Networks, 2657–2663, DOI: https://doi.org/10.1109/IJCNN.2011.6033566 (2011).
  15. Formal Inference From More Than One Model: Multimodel Inference (MMI), 149–205 (Springer New York, New York, NY, 2002).
  16. Jolliffe, I. & Springer-Verlag. Principal Component Analysis. Springer Series in Statistics (Springer, 2002).
  17. Foundations of Data Science (Cambridge University Press, 2020).
  18. Data-driven science and engineering: Machine learning, dynamical systems, and control (Cambridge University Press, 2019).
  19. Prediction via orthogonalized model mixing. \JournalTitleJ. Am. Stat. Assoc. 91, 1197–1208, DOI: https://doi.org/10.1080/01621459.1996.10476989 (1996).
  20. Forecast combination through dimension reduction techniques. \JournalTitleInt. J. Forecast. 27, 224–237, DOI: https://doi.org/10.1016/j.ijforecast.2010.01.012 (2011).
  21. Generalized shrinkage methods for forecasting using many predictors. \JournalTitleJ. Bus. Econ. Stat. 30, 481–493, DOI: https://doi.org/10.1080/07350015.2012.715956 (2012).
  22. Bayesian approach to model-based extrapolation of nuclear observables. \JournalTitlePhysical Review C 98, 034318, DOI: https://doi.org/10.1103/PhysRevC.98.034318 (2018).
  23. The approximation of one matrix by another of lower rank. \JournalTitlePsychometrika 1, 211–218, DOI: https://doi.org/10.1007/BF02288367 (1936).
  24. Gelman, A. et al. Bayesian Data Analysis (CRC Pres, 2013), third edn.
  25. Hoff, P. D. A first course in Bayesian statistical methods, vol. 580 (Springer, 2009).
  26. Weizsäcker, C. F. v. Zur theorie der kernmassen. \JournalTitleZ. Phys. 96, 431–458, DOI: https://doi.org/10.1007/BF01337700 (1935).
  27. The nuclear many-body problem (Springer-Verlag, Berlin, 1980).
  28. From finite nuclei to the nuclear liquid drop: Leptodermous expansion based on self-consistent mean-field theory. \JournalTitlePhys. Rev. C 73, 014309, DOI: https://doi.org/10.1103/PhysRevC.73.014309 (2006).
  29. Neutron drip line in the Ca region from Bayesian model averaging. \JournalTitlePhysical Review Letters 122, 062502, DOI: https://doi.org/10.1103/PhysRevLett.122.062502 (2019).
  30. Neufcourt, L. et al. Quantified limits of the nuclear landscape. \JournalTitlePhys. Rev. C 101, 044307, DOI: https://doi.org/10.1103/PhysRevC.101.044307 (2020).
  31. Uncertainty quantification of mass models using ensemble bayesian model averaging. \JournalTitlePhys. Rev. C 109, 054301, DOI: https://doi.org/10.1103/PhysRevC.109.054301 (2024).
  32. Mass Explorer (2020). http://massexplorer.frib.msu.edu.
  33. HFB-24 mass formula (2020). http://www.astro.ulb.ac.be/bruslib/nucdata/hfb24-dat.
  34. Nuclear ground-state masses and deformations: FRDM(2012). \JournalTitleAt. Data Nucl. Data Tables 109-110, 1 – 204, DOI: https://doi.org/10.1016/j.adt.2015.10.002 (2016).
  35. New relativistic mean-field interaction with density-dependent meson-nucleon couplings. \JournalTitlePhys. Rev. C 71, 024312, DOI: https://doi.org/10.1103/PhysRevC.71.024312 (2005).
  36. Relativistic mean-field interaction with density-dependent meson-nucleon vertices based on microscopical calculations. \JournalTitlePhysical Review C 84, 054309, DOI: https://doi.org/10.1103/physrevc.84.054309 (2011).
  37. Relativistic nuclear energy density functionals: Adjusting parameters to binding energies. \JournalTitlePhysical Review C 78, 034318, DOI: https://doi.org/10.1103/physrevc.78.034318 (2008).
  38. Lalazissis, G. et al. The effective force NL3 revisited. \JournalTitlePhysics Letters B 671, 36–41, DOI: https://doi.org/10.1016/j.physletb.2008.11.070 (2009).
  39. Towards a better parametrisation of Skyrme-like effective forces: a critical study of the SkM force. \JournalTitleNuclear Physics A 386, 79–100, DOI: https://doi.org/10.1016/0375-9474(82)90403-1 (1982).
  40. Hartree-Fock-Bogolyubov description of nuclei near the neutron-drip line. \JournalTitleNucl. Phys. A 422, 103–139, DOI: https://doi.org/10.1016/0375-9474(84)90433-0 (1984).
  41. New Skyrme effective forces for supernovae and neutron rich nuclei. \JournalTitlePhysica Scr. 1995, 231, DOI: https://doi.org/10.1088/0031-8949/1995/T56/034 (1995).
  42. Variations on a theme by Skyrme: A systematic study of adjustments of model parameters. \JournalTitlePhys. Rev. C 79, 034310, DOI: https://doi.org/10.1103/PhysRevC.79.034310 (2009).
  43. Kortelainen, M. et al. Nuclear energy density optimization. \JournalTitlePhys. Rev. C 82, 024313, DOI: https://doi.org/10.1103/PhysRevC.82.024313 (2010).
  44. Kortelainen, M. et al. Nuclear energy density optimization: Large deformations. \JournalTitlePhys. Rev. C 85, 024304, DOI: https://doi.org/10.1103/PhysRevC.85.024304 (2012).
  45. Kortelainen, M. et al. Nuclear energy density optimization: Shell structure. \JournalTitlePhys. Rev. C 89, 054314, DOI: https://doi.org/10.1103/PhysRevC.89.054314 (2014).
  46. New Kohn-Sham density functional based on microscopic nuclear and neutron matter equations of state. \JournalTitlePhys. Rev. C 87, 064305, DOI: https://doi.org/10.1103/PhysRevC.87.064305 (2013).
  47. First Gogny-Hartree-Fock-Bogoliubov nuclear mass model. \JournalTitlePhys. Rev. Lett. 102, 242501, DOI: https://doi.org/10.1103/PhysRevLett.102.242501 (2009).
  48. Probabilistic forecasts, calibration and sharpness. \JournalTitleJ. Roy. Stat. Soc. Ser. B Stat. Methodol. 69, 243–268, DOI: https://doi.org/10.1111/j.1467-9868.2007.00587.x (2007).
  49. Strictly proper scoring rules, prediction, and estimation. \JournalTitleJ. Amer. Statist. Assoc. 102, 359–378, DOI: https://doi.org/10.1198/016214506000001437 (2007).
  50. BMEX - The Bayesian Mass Explorer, DOI: https://doi.org/10.5281/zenodo.7111988 (2022).
  51. Taweret: a Python package for Bayesian model mixing, DOI: https://doi.org/10.48550/arXiv.2310.20549 (2023).
  52. The AME2003 atomic mass evaluation: (II). Tables, graphs and references. \JournalTitleNucl. Phys. A 729, 337 – 676, DOI: https://doi.org/10.1016/j.nuclphysa.2003.11.003 (2003).
  53. The AME 2020 atomic mass evaluation (II). Tables, graphs and references. \JournalTitleChin. Phys. C 45, 030003, DOI: https://doi.org/10.1088/1674-1137/abddaf (2021).
Citations (2)
View on Semantic Scholar

Summary

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

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now