doi abstract bibtex

Scenarios have been discovered in which it is impossible to prove whether or not a machine-learning algorithm could solve a particular problem. This finding might have implications for both established and future learning algorithms. [Excerpt] During the twentieth century, discoveries in mathematical logic revolutionized our understanding of the very foundations of mathematics. In 1931, the logician Kurt Gödel showed that, in any system of axioms that is expressive enough to model arithmetic, some true statements will be unprovable [...] Writing in Nature Machine Intelligence, Ben-David et al. show that the field of machine learning, although seemingly distant from mathematical logic, shares this limitation. They identify a machine-learning problem whose fate depends on the continuum hypothesis, leaving its resolution forever beyond reach. [...] [] [...] [...] Machine learning has matured as a mathematical discipline and now joins the many subfields of mathematics that deal with the burden of unprovability and the unease that comes with it. Perhaps results such as this one will bring to the field of machine learning a healthy dose of humility, even as machine-learning algorithms continue to revolutionize the world around us.

@article{reyzinUnprovabilityComesMachine2019, title = {Unprovability Comes to Machine Learning}, author = {Reyzin, Lev}, year = {2019}, month = jan, volume = {565}, pages = {166--167}, issn = {0028-0836}, doi = {10.1038/d41586-019-00012-4}, abstract = {Scenarios have been discovered in which it is impossible to prove whether or not a machine-learning algorithm could solve a particular problem. This finding might have implications for both established and future learning algorithms. [Excerpt] During the twentieth century, discoveries in mathematical logic revolutionized our understanding of the very foundations of mathematics. In 1931, the logician Kurt G\"odel showed that, in any system of axioms that is expressive enough to model arithmetic, some true statements will be unprovable [...] Writing in Nature Machine Intelligence, Ben-David et al. show that the field of machine learning, although seemingly distant from mathematical logic, shares this limitation. They identify a machine-learning problem whose fate depends on the continuum hypothesis, leaving its resolution forever beyond reach. [...] [] [...] [...] Machine learning has matured as a mathematical discipline and now joins the many subfields of mathematics that deal with the burden of unprovability and the unease that comes with it. Perhaps results such as this one will bring to the field of machine learning a healthy dose of humility, even as machine-learning algorithms continue to revolutionize the world around us.}, journal = {Nature}, keywords = {*imported-from-citeulike-INRMM,~INRMM-MiD:c-14678591,artificial-intelligence,epistemology,limiting-factor,machine-learning,mathematics,modelling-uncertainty,software-uncertainty,unexpected-effect}, lccn = {INRMM-MiD:c-14678591}, number = {7738} }

Downloads: 0