Abstract
This chapter provides an analysis of the relationship of the traditional problems of justifying inductive inferences to the field of machine learning. After the summary of the philosophical problems of induction, text focus on the two philosophical problems relevant to the supervised and unsupervised machine learning. The former is a famous new riddle of induction devised by N. Goodman. The author argues that remarkable results in the theory of machine learning, no-free-lunch theorems are a formalisation of this traditional philosophical problem. Consequently, lengthy philosophical discussions on this problem are relevant to these results and vice versa. The later problem is the problem of similarity, as identified by N. Goodman and W. V. Quine. It is claimed that those discussions can help practitioners of unsupervised learning to be aware of its limitations.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
This part of the text is an extended version of the author’s text “How Gruesome are the No-free-lunch Theorems for Machine Learning?” [16].
- 2.
We suppose that the new riddle is a different issue from the classical problem of induction, what is the received position with a few notable exception.
- 3.
Rodriguez-Pereyra claims there are strong philosophical grounds that Leibniz thought that the similarity of substances does not derive from the similarity of their properties (accidents) [24].
- 4.
Although there are readings of Hume that suggest that Hume’s position on similarity is closer to Goodman’s view [7].
References
Blackburn S (1984) Spreading the word. Oxford University Press
Bacon F, Thomas F (1889) Novum organum. Clarendon Press, Oxford
Cha S-H (2007) Comprehensive survey on distance/similarity measures between probability density functions. City 1(2):1
Domingos P (2012) A few useful things to know about machine learning. Commun ACM 55(10):78–87
Domingos P (2015) The master algorithm: how the quest for the ultimate learning machine will remake our world. Hachette, UK
Dretske F, Carnap R, Bar-Hillel Y (1953) Semantic theories of information. Philos Sci 4(14):147–157
Gamboa S (2007) Hume on resemblance, relevance, and representation. Hume Stud 33/1:21–40
Giraud-Carrier C, Provost F (2005) Toward a justification of meta-learning: is the no free lunch theorem a show-stopper. In: Proceedings of the ICML-2005 workshop on meta-learning
Goodman N (1946) A query on confirmation. J Philos 43(14):383–385
Goodman N (1971) Problems and projects. Bobbs-Merrill
Goodman N (1983) Fact, fiction, and forecast. Harvard University Press
Hume D (1748) An inquiry concerning human understanding. Clarendon Press
Igel C, Toussaint M (2005) A no-free-lunch theorem for non-uniform distributions of target functions. J Math Model Algorithms 3(4):313–322
Joyce T, Herrmann JM (2018) A review of no free lunch theorems, and their implications for metaheuristic optimisation. Stud Comput Intell 744:27–51
Kubat M (2017) Introduction to machine learning. Springer International Publishing
Lauc D (2018) How gruesome are the no-free-lunch theorems for machine learning? Croat J Philos 18(54):479–489
Lauc D (2019) Reasoning about inexact dates using dense vector representation. Compusoft 8:3031–3035
Leibniz GW, Ritter P (1950) Sämtliche schriften und briefe. Akademie-Verlag
Nielsen F (2010) A family of statistical symmetric divergences based on Jensen’s inequality. arXiv:1009.4004
Orenstein A, Kotatko P (2012) Knowledge, language and logic: questions for Quine. Springer Science & Business Media
Pakaluk M (1989) Quine’s 1946 lecture on Hume. J Hist Philos 445–459
Popper K (2002) Conjectures and refutations: the growth of scientific knowledge. Routledge
Quine WVO (1970) Natural kinds. D. Reidel
Rodriguez-Pereyra G (2014) Leibniz’s principle of identity of indiscernibles. OUP
Schaffer C (1994) A conservation law for generalization performance. Mach Learn Proc 1994:259–265
Tversky A (1977) Features of similarity. Psychol Rev 84:327–354
Valiant L (2013) Probably approximately correct: nature’s algorithms for learning and prospering in a complex world. Hachette
Wolpert D (1992) Stacked generalization. Neural Netw 5:241–259
Wolpert D (1996) The lack of a priori distinctions between learning algorithms. Neural Comput 1341–1390
Wolpert D (2013) Ubiquity symposium: evolutionary computation and the processes of life: what the no free lunch theorems really mean: how to improve search algorithms. Ubiquity 2
Wolpert D, Macready WG (1995) No free lunch theorems for search. Technical report SFI-TR-95-02-010, Santa Fe Institute
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Lauc, D. (2020). Machine Learning and the Philosophical Problems of Induction. In: Skansi, S. (eds) Guide to Deep Learning Basics. Springer, Cham. https://doi.org/10.1007/978-3-030-37591-1_9
Download citation
DOI: https://doi.org/10.1007/978-3-030-37591-1_9
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-37590-4
Online ISBN: 978-3-030-37591-1
eBook Packages: Computer ScienceComputer Science (R0)