Long-Term Evolution Experiment with Genetic Programming

Langdon, William B.; Banzhaf, Wolfgang

doi:10.1162/artl_a_00360

Cited by 7 publications

(6 citation statements)

References 45 publications

(55 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We have used various recent efficiency improvements to Singleton's GPquick [8][9][10]. In particular with extended runs [17] incremental evaluation [11] again gave substantial speed up without affecting evolution.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Failed disruption propagation in integer genetic programming

Langdon

2022

Proceedings of the Genetic and Evolutionary Computation Conference Companion

Self Cite

View full text Add to dashboard Cite

We inject a random value into the evaluation of highly evolved deep integer GP trees 9 743 720 times and find 99.7% of test outputs are unchanged. Suggesting crossover and mutation's impact are dissipated and seldom propagate outside the program. Indeed only errors near the root node have impact and disruption falls exponentially with depth at between e −depth/3 and e −depth/5 for recursive Fibonacci GP trees, allowing five to seven levels of nesting between the runtime perturbation and an optimal test oracle for it to detect most errors. Information theory explains this locally flat fitness landscape is due to FDP. Overflow is not important and instead, integer GP, like deep symbolic regression floating point GP and software in general, is not fragile, is robust, is not chaotic and suffers little from Lorenz' butterfly.

show abstract

Section: Discussionmentioning

confidence: 99%

“…Figure2: Colour shows locations where run time disruption impacts output of a large evolved GP tree. Brighter colours mean more test cases(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). Tree plotted with output at center of circular lattice[3].…”

mentioning

confidence: 99%

Failed disruption propagation in integer genetic programming

Langdon

2022

Proceedings of the Genetic and Evolutionary Computation Conference Companion

Self Cite

View full text Add to dashboard Cite

show abstract

“…These models are followed by specialised discussions of their implications. Section 5.5 suggests when there is no practical limit on tree growth in extended evolution in floating point GP [28]. Whilst Section 5.6 looks at other function sets and the impact of side effects in traditional imperative programming.…”

Section: Discussionmentioning

confidence: 99%

“…2] (albeit using a different function set). In [28,Fig. 2] we reported bloat to trees of more than 64 million nodes.…”

Section: Failed Disruption Propagation As a Limit To Gp Tree Bloatmentioning

confidence: 99%

Deep Genetic Programming Trees Are Robust

Langdon

2022

ACM Trans. Evol. Learn. Optim.

Self Cite

View full text Add to dashboard Cite

We sample the genetic programming tree search space and show it is smooth since many mutations on many test cases have little or no fitness impact. We generate uniformly at random high order polynomials composed of 12 500 and 750 000 additions and multiplications and follow the impact of small changes to them. From information theory 32 bit floating point arithmetic is dissipative and even with 1501 test cases deep mutations seldom have any impact on fitness. Absolute difference between parent and child evaluation can grow as well as fall further from the code change location but the number of disrupted fitness tests falls monotonically. In many cases deeply nested expressions are robust to crossover syntax changes, bugs, errors, run time glitches, perturbations, etc., because their disruption falls to zero, and so it fails to propagate beyond the program.

show abstract

“…The study compares the performance of learning using a GPU and a CPU with AVX. The authors of [13] experimented with AVX for genetic programming, especially with generations of binary trees. AVX allowed them to execute faster and far longer than previous attempts.…”

Section: Introductionmentioning

confidence: 99%

Acceleration of Particle Swarm Optimization with AVX Instructions

Šafařík

Snášel

2023

Applied Sciences

View full text Add to dashboard Cite

Parallel implementations of algorithms are usually compared with single-core CPU performance. The advantage of multicore vector processors decreases the performance gap between GPU and CPU computation, as shown in many recent pieces of research. With the AVX-512 instruction set, there will be another performance boost for CPU computations. The availability of parallel code running on CPUs made them much easier and more accessible than GPUs. This article compares the performances of parallel implementations of the particle swarm optimization algorithm. The code was written in C++, and we used various techniques to obtain parallel execution through Advanced Vector Extensions. We present the performance on various benchmark functions and different problem configurations. The article describes and compares the performance boost gained from parallel execution on CPU, along with advantages and disadvantages of parallelization techniques.

show abstract

Long-Term Evolution Experiment with Genetic Programming

Cited by 7 publications

References 45 publications

Failed disruption propagation in integer genetic programming

Failed disruption propagation in integer genetic programming

Deep Genetic Programming Trees Are Robust

Acceleration of Particle Swarm Optimization with AVX Instructions

Contact Info

Product

Resources

About