BPadvertisementfrom

  • Subscribe to our RSS feed.
  • Twitter
  • StumbleUpon
  • Reddit
  • Facebook
  • Digg

Friday, 28 June 2013

Solomonoff Universal Induction

Posted on 19:00 by Unknown
In an earlier post (Kolmogorov, Solomonoff, and de Finetti) I linked to a historical article on the problem of induction. Here's an even better one, which gives a very clear introduction to Solomonoff Induction.
A Philosophical Treatise of Universal Induction
Samuel Rathmanner, Marcus Hutter

Understanding inductive reasoning is a problem that has engaged mankind for thousands of years. This problem is relevant to a wide range of fields and is integral to the philosophy of science. It has been tackled by many great minds ranging from philosophers to scientists to mathematicians, and more recently computer scientists. In this article we argue the case for Solomonoff Induction, a formal inductive framework which combines algorithmic information theory with the Bayesian framework. Although it achieves excellent theoretical results and is based on solid philosophical foundations, the requisite technical knowledge necessary for understanding this framework has caused it to remain largely unknown and unappreciated in the wider scientific community. The main contribution of this article is to convey Solomonoff induction and its related concepts in a generally accessible form with the aim of bridging this current technical gap. In the process we examine the major historical contributions that have led to the formulation of Solomonoff Induction as well as criticisms of Solomonoff and induction in general. In particular we examine how Solomonoff induction addresses many issues that have plagued other inductive systems, such as the black ravens paradox and the confirmation problem, and compare this approach with other recent approaches.
Of course, properties such as Turing machine-independence and other key results are asymptotic in nature (only in the limit of very long sequences of data does it cease to matter exactly which reference Turing machine you choose to define program length). When it comes to practical implementations, the devil is in the details! You can think of the Solomonoff Universal Prior as a formalization of  the a priori assumption that the information in our Universe is highly compressible (i.e., there are underlying simple algorithms -- laws of physics -- governing its evolution). See also Information, information processing and black holes. From the paper:
... The formalization of Solomonoff induction makes use of concepts and results from computer science, statistics, information theory, and philosophy. It is interesting that the development of a rigorous formalization of induction, which is fundamental to almost all scientific inquiry, is a highly multi-disciplinary undertaking, drawing from these various areas. Unfortunately this means that a high level of technical knowledge from these various disciplines is necessary to fully understand the technical content of Solomonoff induction. This has restricted a deep understanding of the concept to a fairly small proportion of academia which has hindered its discussion and hence progress.

... Every major contribution to the foundations of inductive reasoning has been a contribution to under- standing rational thought. Occam explicitly stated our natural disposition towards simplicity and elegance. Bayes inspired the school of Bayesianism which has made us much more aware of the mechanics behind our belief system. Now, through Solomonoff, it can be argued that the problem of formalizing optimal inductive inference is solved.

Being able to precisely formulate the process of (universal) inductive inference is also hugely significant for general artificial intelligence. Obviously reasoning is synonymous with intelligence, but true intelligence is a theory of how to act on the conclusions we make through reasoning. It may be argued that optimal intelligence is nothing more than optimal inductive inference combined with optimal decision making. Since Solomonoff provides optimal inductive inference and decision theory solves the problem of choosing optimal actions, they can therefore be combined to produce intelligence. ... [ Do we really need Solomonoff? Did Nature make use of his Universal Prior in producing us? It seems like cheaper tricks can produce "intelligence" ;-) ]
Here are some nice informal comments by Solomonoff himself.
Email ThisBlogThis!Share to XShare to FacebookShare to Pinterest
Posted in ai, machine learning, mathematics, probability, statistics | No comments
Newer Post Older Post Home

0 comments:

Post a Comment

Subscribe to: Post Comments (Atom)

Popular Posts

  • PhD Comics: the movie
    PHD Movie Trailer from PHD Comics on Vimeo . I met Jorge Cham , the cartoonist who draws PhD Comics, a few years ago at Sci Foo. Cham was ...
  • Finding the Next Einstein
    Duke researcher Jonathan Wai interviewed me for his Psychology Today blog, Finding the Next Einstein . Below are my answers to two of his q...
  • Beanbags and causal variants
    Not only do these results implicate common causal variants as the source of heritability in disease susceptibility, but they also suggest th...
  • Sitzfleisch
    Freeman Dyson reviews the new biography of Oppenheimer by Ray Monk. I discussed the book already here . NYBooks : ... The subtitle, “A Life ...
  • UC Davis colloquium
    I'll be giving a colloquium at UC Davis on Monday. Please come if you can! slides Title: Genetics, intelligence and other quantitati...
  • A blog is born
    Raghu Parasarathy , a biophysicist at U Oregon, and my correspondent in this previous post on faculty blogging, has decided to try it out. ...
  • News from Microsoft Research Faculty Summit 2013
    Measuring the maximal commuting subset of observables uniquely determines the pure state of a quantum system (recently proved Kadison-Singer...
  • Talk cancelled
    This talk has been cancelled, for complex reasons that I will not discuss.
  • What is the difference?
    Some excerpts from a talk by Richard Hamming (inventor of, among other things, hamming codes ) on how to do great research. I recommend rea...
  • East Asian sociopaths?
    Some would assert that CEOs and other people in leadership positions are often warm sociopaths . Interestingly, it is claimed that there is ...

Categories

  • ability (2)
  • academia (9)
  • affirmative action (8)
  • ai (13)
  • aig (1)
  • alan turing (3)
  • algorithms (2)
  • alpha (2)
  • american society (54)
  • art (3)
  • ashkenazim (1)
  • aspergers (4)
  • athletics (6)
  • autism (4)
  • autobiographical (13)
  • basketball (4)
  • bayes (1)
  • behavioral economics (4)
  • berkeley (5)
  • bgi (24)
  • biology (23)
  • biotech (6)
  • bjj (5)
  • black holes (4)
  • blade runner (2)
  • blogging (3)
  • books (5)
  • borges (2)
  • bounded rationality (10)
  • brainpower (57)
  • bubbles (3)
  • caltech (14)
  • cambridge uk (1)
  • careers (18)
  • charles darwin (1)
  • chet baker (2)
  • China (25)
  • christmas (1)
  • class (2)
  • cognitive science (35)
  • cold war (1)
  • complexity (1)
  • computing (9)
  • conferences (4)
  • cosmology (4)
  • creativity (3)
  • credit crisis (10)
  • crossfit (5)
  • cryptography (2)
  • data mining (4)
  • dating (2)
  • demographics (1)
  • derivatives (5)
  • determinism (1)
  • digital books (1)
  • dna (4)
  • economic history (5)
  • economics (38)
  • econtalk (2)
  • ecosystems (1)
  • education (5)
  • efficient markets (8)
  • Einstein (2)
  • elitism (14)
  • encryption (1)
  • energy (1)
  • entrepreneurs (3)
  • entropy (1)
  • environmentalism (1)
  • eugene (3)
  • evolution (19)
  • expert prediction (6)
  • fake alpha (2)
  • feminism (2)
  • Fermi problems (2)
  • feynman (7)
  • film (9)
  • finance (42)
  • fitness (3)
  • flynn effect (1)
  • foo camp (1)
  • football (5)
  • france (1)
  • free will (1)
  • freeman dyson (2)
  • fx (2)
  • game theory (1)
  • geeks (2)
  • gender (4)
  • genetic engineering (15)
  • genetics (79)
  • genius (24)
  • genomics (2)
  • geopolitics (7)
  • gilded age (13)
  • global warming (1)
  • globalization (23)
  • godel (2)
  • goldman sachs (2)
  • google (4)
  • happiness (2)
  • harvard (8)
  • harvard society of fellows (5)
  • hedge funds (4)
  • hedonic treadmill (1)
  • height (2)
  • higher education (38)
  • history (8)
  • history of science (12)
  • hormones (3)
  • hugh everett (2)
  • human capital (34)
  • humor (1)
  • income inequality (21)
  • india (2)
  • industrial revolution (1)
  • innovation (38)
  • intellectual history (10)
  • intellectual property (1)
  • intellectual ventures (1)
  • internet (4)
  • iq (16)
  • italy (4)
  • james salter (3)
  • japan (4)
  • jiujitsu (8)
  • keynes (1)
  • kids (13)
  • lewontin fallacy (1)
  • lhc (1)
  • literature (12)
  • luck (1)
  • machine learning (8)
  • malcolm gladwell (1)
  • manhattan (2)
  • many worlds (10)
  • mathematics (14)
  • meritocracy (7)
  • microsoft (2)
  • mma (10)
  • monsters (2)
  • moore's law (1)
  • movies (9)
  • MSU (18)
  • music (5)
  • mutants (2)
  • nathan myhrvold (1)
  • neal stephenson (1)
  • neanderthals (2)
  • nerds (3)
  • net worth (5)
  • neuroscience (7)
  • new yorker (1)
  • nicholas metropolis (1)
  • noam chomsky (2)
  • nobel prize (2)
  • nsa (2)
  • nuclear weapons (5)
  • obama (7)
  • olympics (4)
  • oppenheimer (7)
  • patents (1)
  • personality (9)
  • philip k. dick (1)
  • philosophy of mind (2)
  • photos (40)
  • physical training (13)
  • physics (73)
  • podcasts (10)
  • political correctness (6)
  • politics (4)
  • pop culture (2)
  • prisoner's dilemma (1)
  • privacy (2)
  • probability (5)
  • prostitution (2)
  • psychology (25)
  • psychometrics (31)
  • qcd (1)
  • quants (9)
  • quantum computers (2)
  • quantum field theory (3)
  • quantum mechanics (18)
  • race relations (10)
  • real estate (1)
  • realpolitik (6)
  • renaissance technologies (1)
  • research (3)
  • russia (2)
  • sad but true (2)
  • sci fi (8)
  • science (42)
  • sec (1)
  • security (5)
  • silicon valley (6)
  • singularity (1)
  • smpy (1)
  • social networks (2)
  • social science (12)
  • software development (2)
  • solar energy (1)
  • sports (13)
  • startups (19)
  • statistics (16)
  • success (2)
  • taiwan (1)
  • talks (16)
  • teaching (2)
  • technology (34)
  • television (2)
  • travel (24)
  • turing test (1)
  • ufc (8)
  • ultimate fighting (1)
  • universities (33)
  • university of oregon (6)
  • usain bolt (2)
  • venture capital (3)
  • volatility (1)
  • von Neumann (10)
  • wall street (2)
  • war (1)
  • warren buffet (1)
  • wwii (3)

Blog Archive

  • ▼  2013 (134)
    • ►  August (10)
    • ►  July (15)
    • ▼  June (22)
      • History will remember their names
      • Bert and Ernie, Out?
      • Solomonoff Universal Induction
      • Views of Hong Kong
      • Kolmogorov, Solomonoff, and de Finetti
      • HDL-C heritability from whole genomes: common vari...
      • Android Dreams
      • WDIST and PLINK
      • Ken Wilson, dead at 77
      • Beanbags and causal variants
      • Sinica podcast
      • China 3.0
      • Erdos and Tao
      • Spy vs Spy
      • The ratchet of power
      • If you can't fix it you've got to stand it
      • Horizons of truth
      • Weak Meat Strong Eat
      • Morgan Freeman on physics and physicists
      • All that is left is the wind in the pines
      • Lore
      • The genetics of humanness
    • ►  May (20)
    • ►  April (21)
    • ►  March (18)
    • ►  February (14)
    • ►  January (14)
  • ►  2012 (222)
    • ►  December (17)
    • ►  November (19)
    • ►  October (20)
    • ►  September (25)
    • ►  August (19)
    • ►  July (18)
    • ►  June (16)
    • ►  May (20)
    • ►  April (16)
    • ►  March (18)
    • ►  February (20)
    • ►  January (14)
  • ►  2011 (144)
    • ►  December (20)
    • ►  November (16)
    • ►  October (25)
    • ►  September (23)
    • ►  August (21)
    • ►  July (26)
    • ►  June (13)
Powered by Blogger.

About Me

Unknown
View my complete profile