Research Article

A Computable Measure of Algorithmic Probability by Finite Approximations with an Application to Integer Sequences

Figure 2

(a) Correlation between the estimated algorithmic complexity () by the AP-based measure (BDM) and the length of the text description of each sequence from the OEIS. Fitted line for highest correlation (BDM) is given by using least squares. (b) Algorithmic complexity estimation by BDM () and of compression on program length (in the Wolfram Language/Mathematica) as coming from the OEIS. In parenthesis, the Spearman rank correlation values for each case. Further compressing the program length using "compress" resulted in a lower correlation value and BDM outperformed lossless compression.
(a)
(b)