Redundancy and computational efficiency in Cartesian genetic programming

By Julian Francis Miller, Stephen Leslie Smith. 2006

294 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Molecular Biology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Molecular Biology

Citations in the corpus

AutoML: A Survey of the State-of-the-Art (2019, methodology): “To represent variable-length network structures, Suganuma et al. [21] use Cartesian genetic programming (CGP) [101], [102] encoding scheme to represent the network as directed acyclic graphs, for the CNN architecture representation.”
A genetic programming approach to designing convolutional neural network architectures (2017, methodology): “Moreover, to maintain a neutral dri, which is eective for CGP evolution [21, 22], we modify a parent by the neutral mutation if”
Evolving simple programs for playing atari games (2018, background): “While these nodes are not used in the nal program, they have been shown to aid evolutionary search [15, 28, 30].”
Neuro-Centric and Holocentric Approaches to the Evolution of Developmental Neural Networks (2014, background): “Miller and Smith showed that in genotypes having 4,000 nodes, the percentage of inactive nodes is approximately 95 %! [47].”
A Field Guide to Genetic Programming (2008, background): “In Miller’s Cartesian GP (Miller, 1999; Miller and Smith, 2006), programs are represented by linear chromosomes containing integers.”
Exploiting the Potential of Standard Convolutional Autoencoders for Image Restoration by Evolutionary Search (2018, unknown): “Moreover, if the ﬁtness values of the children do not improve, then we modify a parent (Miller & Thomson, 2000; Miller & Smith, 2006); in this case, we change only the non-functioning nodes so that the realized CAE (i.e., functioning nodes) will not change (line 14).”
Cartesian genetic programming encoded artificial neural networks: a comparison using three benchmarks (2013, background): “CGP also has many advantages over traditional GP, including redundancy in the chromosomes [29] and natural resilience to bloat [27].”
Developmental Plasticity in Cartesian Genetic Programming based Neural Networks (2016, methodology): “It is observed in the CGP literature that only upto 5% nodes are usually active at runtime, thus having 95% of garbage space (inactive nodes) that may be activated during evolution (Miller and Smith, 2006).”
Evolution of Cartesian Genetic Programs for Development of Learning Neural Architecture (2011, background): “Indeed, it has been shown that evolutionary search proceeds most quickly when extraordinary levels of redundancy are present (i.e., when 95% of all genes are redundant; Miller and Smith, 2006 ).”