90 B. Hodjat and H. Shahrzad
Training continues until the maturity age for all of itsMfolds is reached. At this
point validation can then begin on the remainingNMsegments, by sending the
candidate back down to Evolution Engines designated to sub-segment pair unseen
by the candidate so far.
Note thatMis really a maximum for a given data set. An individual Evolution
Engine can be assigned any number of folds from 1 toM. By doing so, we have
many more permutations of the data sub-segments assigned as originating segments,
and the segments have overlaps. For instance, rather than dividing the data set to a
fixed 4 segments, we can divide it to 8 sub-segments, and assign each Evolution
Engine two of the possible permutations of the sub-segments, which would be a
total of 8 7 D 56 possible pairs to be used as originating segments. This would
mean, however, that a candidate should be prevented from being sent for evaluation
to an Evolution Engine with a combination of sub-segments that include any sub-
segment in the candidate’s originating segment set.
More experimental and theoretical work on the approach and the best settings for
nare also in order.
AcknowledgementsThe authors wish to thank Sentient Technologies for sponsoring this research
and providing the processing capacity required for the experiments presented in this paper.
References
Goldberger AL, Amaral LA, Glass L, Hausdorff JM, Ivanov PC, Mark RG, Mietus JE, Moody GB,
Peng CK, Stanley HE (2000) Physiobank, physiotoolkit, and physionet components of a new
research resource for complex physiologic signals. Circulation 101(23):e215–e220
Hodjat B, Shahrzad H (2013) Introducing an age-varying fitness estimation function. In: Genetic
programming theory and practice X. Springer, Berlin, pp 59–71
Hodjat B, Hemberg E, Shahrzad H, O’Reilly UM (2014) Maintenance of a long running distributed
genetic programming system for solving problems requiring big data. In: Genetic programming
theory and practice XI. Springer, Berlin, pp 65–83
Hornby, GS (2006) ALPS: the age-layered population structure for reducing the problem of pre-
mature convergence. In: Proceedings of the 8th annual conference on Genetic and evolutionary
computation, ACM, New York, pp 815–822
O’Reilly UM, Wagy M, Hodjat B (2013) Ec-star: a massive-scale, hub and spoke, distributed
genetic programming system. In: Genetic programming theory and practice X. Springer, Berlin,
pp 73–85
Refaeilzadeh P, Tang L, Liu H (2009) Cross-validation. In: Encyclopedia of database systems.
Springer, Berlin, pp 532–538
Rivest RL (1987) Learning decision lists. Mach Learn 2(3):229–246