Difference between revisions
problem-of-induction
Status and infobox
- status:
—→ open - kind:
—→ problem - year:
—→ 1739 - statement:
—→ What justifies believing that the future will resemble the past?
Source +18 −0
+---
+title: The problem of induction
+kind: problem
+status: open
+year: 1739
+authors: David Hume
+statement: What justifies believing that the future will resemble the past?
+space: knowledge
+---
+Every argument for induction seems to rely on induction. Popper answered that science does not need induction, only falsification. Bayesian and algorithmic approaches, such as Solomonoff induction, give formal accounts of learning from data, yet each rests on a prior that is itself not justified from experience. The problem is newly practical: machine learning is induction at industrial scale.
+
+Posed 1739, David Hume.
+
+## Sources
+
+- [Wikipedia: Problem of induction](https://en.wikipedia.org/wiki/Problem_of_induction)
+
+[[Space:knowledge|Knowledge]] · [[Continent:philosophy|Philosophy]]
Revision 222 as rendered
Every argument for induction seems to rely on induction. Popper answered that science does not need induction, only falsification. Bayesian and algorithmic approaches, such as Solomonoff induction, give formal accounts of learning from data, yet each rests on a prior that is itself not justified from experience. The problem is newly practical: machine learning is induction at industrial scale.
Posed 1739, David Hume.