Glossary of Terms



 
 
 
Allele Allele, in biology, is the term given to the appropriate range of values for genes. In genetic algorithms, an allele is the value of the gene (or genes).
Allele Loss Allele loss is the natural loss of traits in the gene pool over the generations of a run. Another term for allele loss is convergence. Severe allele loss results in a population incapable of solving the problem with the available gene pool.
Atom Atoms are the mathematical or symbolic operators in the parse tree. Atoms are always internal (non-leaf) nodes as all atoms accept arguments.
Closure Closure is based on the design of genetic operators to ensure the validity of functions generated by the genetic program. If closure is achieved, the functions will not cause errors, regardless of their arrangement, when their fitness is tested.
Convergence Convergence is a reduction in the diversity of genes available in the population due to fitness proportionate operators.
Diversity Diversity is the term used to describe the relative uniqueness of each individual in the population. This condition is considered favourable as the greater the variety of genes available to the genetic algorithm the greater the likelihood of the system identifying alternate solutions.
Function Set The set of functions (atoms) available to the genetic program. The operators in the function set are used to make up the internal nodes of the parse tree.
Generation A generation is an iteration of the genetic algorithm. Conventionally, the initial random generation is known as generation zero.
Generational Equivalent Generational equivalent is a term used in steady state techniques to identify a generation has occurred. Because the populations of two generation are combined during the selection and breeding cycle, a generational equivalent is said to occur when the number of genetic operations is equal to the population size of the genetic program.
Generational GP Generational genetic programming is the process of producing distinct generations in each iteration of the genetic algorithm.
Genotype The genotype is the structure of the solution produced by the genetic program.
Inversion Inversion is a genetic operator sometimes used in genetic algorithms where two points in the string are chosen and the order of the genes between those two points are reversed (inverted).
Leaf Node A leaf node is a node in a parse tree with no children; a terminal.
Non-leaf Node  A non-leaf node is a node in a parse tree with one or more children; a function.
Non-Linear problems Non-linear problems are those problems where the relationship between the inputs and the outputs of the problem are not clearly discernible.
NP-complete Non-Deterministic Polynomial Complete is a term used in complexity theory to identify a particular class of problem. In an NP-complete problem, the relationship between the number of input parameters to the problem and the problem complexity is exponential. If an enumerative search strategy is adopted, this exponential increase in problem complexity results in an exponential increase in the time to solve the problem.
Off-line  Off-line learning is a learning method where the systems learning is conducted prior to its practical application. When the system is actually used, no learning takes place.
On-line On-line learning is a learning method where the system learns while it is being used. This learning method is more robust than off-line learning as the system can adapt to changes that may occur in the systems environment.
Parse Tree A parse tree is the way the genetic programming paradigm represents the functions generated by a genetic program. A parse tree is similar to a binary decision tree and preorder traversal of the tree produces an S-expression that represents a potential solution in reverse polish notation.
Parsimony Parsimony is the simplicity of the structure of a function.
Recombination Recombination is another name for crossover.
Reproduction Reproduction is the copying of a chromosome into the next generation, i.e. budding.
S-expression  S-expressions are more frequently called LISP expressions and represent a function and its arguments. S-expressions are the most common representation of functions generated in genetic programming.
Structural Complexity  Structural complexity is the term used to describe the number of nodes in the parse tree, i.e. Number of Terminals + Number of Atoms = Structural Complexity.
Sufficiency Sufficiency refers to a necessary variety of atoms and terminals available to the genetic program to solve the problem.
Terminals Terminals are the numeric values, (variables, constants and zero argument functions) in the parse tree and are always external (leaf) nodes in the tree. The terminals act as arguments for the operator (atom) that is their parent in the tree.
Terminal Set The terminal set is comprised of the terminals available to the genetic program.
Tuning Tuning is the process of selecting the appropriate genetic operators and their respective parameters to suit a problem.
Wrapper A wrapper is an interpretative function that evaluates the expression to be tested and returns a value within a suitable range for the simulation.

 

[Chapter 7] [Bibliography]
[Beer]