A CODING METHOD FOR INDUCTIVE INFERENCE

Abstract

A new general inductive inference method has been described in which the a-priori probability of a sequence of symbols is computed on the basis of the lengths of various code strings that could be used to describe that sequence to a universal Turing machine. A coding method is displayed for a simple Bernoulli sequence and the inference technique is applied to the computation of probabilities of symbols in that sequence. The results obtained in this case are shown to be identical to Laplace's rule of succession. The probabilities correspond to Shannon's entropy if the Bernoulli sequence is a very long one. (Author)

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 1961
Accession Number
AD0256886

Entities

People

  • R.j. Solomonoff

Tags

DTIC Thesaurus Topics

  • Automata
  • Computations
  • Machines
  • Probability
  • Sequences

Fields of Study

  • Mathematics

Readers

  • Computational Linguistics
  • Computer Science/Computer Engineering/Data Science/Digital Signal Processing.
  • Radio communications and signal processing.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Machine Learning Algorithms