A program searching for a functional dependence using genetic programming with coefficient adjustment

Hlaváč, Vladimír

doi:10.1109/scsp.2016.7501014

Cited by 4 publications

(4 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results demonstrate the ability of stochastic and deterministic method of constant evaluation to solve proposed problem. In the case of tested functions, the deterministic method of the gradient descent, proposed in [7], was faster (see Fig. 9).…”

Section: Resultsmentioning

confidence: 97%

“…All the tested methods were implemented into the program described in [7]. In this program, a slightly different method of constant implementation is used; instead of a leaf object in the tree of the represented function, a point where a multiplicative constant is used is only marked in the tree.…”

Section: The Programmentioning

confidence: 99%

“…This is a native setting of the used application, where the teaching coefficient γ of 0.3 (ratio 1.3) is used, but restarted five times from different points of the geometric sequence. This setting was part of the development of this application [7].…”

Section: Gradient Descentmentioning

confidence: 99%

“…A more straightforward approach is described in [7], where a gradient method is implemented. Because of a different definition of the problem, the Exponentiated Gradient Descent [9] method was used, alternatively with or without a variable step rate, known from the Simulated Annealing [8].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Genetic Programming With Either Stochastic or Deterministic Constant Evaluation

Hlaváč¹

2018

NNW

View full text Add to dashboard Cite

Constant evaluation is a key problem for symbolic regression, one solved by means of genetic programming. For constant evaluation, other evolutionary methods are often used. Typical examples are some variants of genetic programming or evolutionary systems, all of which are stochastic. The article compares these methods with a deterministic approach using exponentiated gradient descent. All the methods were tested on single sample function to maintain the same conditions and results are presented in graphs. Finally, three different tasks (ten times each) are compared to check the reliability of the methods tested in the article.

show abstract

Section: Resultsmentioning

confidence: 97%

Section: The Programmentioning

confidence: 99%

Section: Gradient Descentmentioning

confidence: 99%