EpochX

Source code of the modified EpochX (version 1.4.1) framework for the Automated Multi-Label Classification (Auto-MLC) project. The original code is available at https://www.epochx.org/ and https://github.com/tc33/EpochX.

Main differences to the original version:

1. It allows the generation of random float numbers (from a range) into the production rules, by using "RANDFLOAT(0.0,1.0)";
2. It allows the generation of random integer numbers (from a range) into the production rules, by using "RANDINT_TYPE0(1,64)";
3. It allows the generation of random integer numbers (from a range) into the production rules, restricted by the square root of the number of class labels plus one, by using "RANDINT_TYPE1(1,SQRT(L)+1)";
4. It allows the generation of random integer numbers (from a range) into the production rules, restricted by the number of class labels divided by two, by using "RANDINT_TYPE2(1,L/2)";
5. It allows the generation of random integer numbers (from a range) into the production rules, restricted by the minimum value between 100 and two times the number of class labels, by using "RANDINT_TYPE3(2,min(2L;100))";
6. It allows the generation of random integer numbers (from a range) into the production rules, restricted by the range from 20% to 100% of the number of attributes of the problem, by using "RANDINT_TYPE4((0.2,1.0)*n_attributes)";
7. It allows the generation of random integer numbers (from a range) into the production rules, restricted by the number of class labels of the problem, by using "RANDINT_TYPE5(1,L)";
8. It allows the generation of random integer numbers (from a range) into the production rules, restricted by the number of class labels minus one (1), by using "RANDINT_TYPE6(2,L-1)";
9. Because of legibility issues of the grammar, we automatically add space to separate the produced terminals. Hence, the user that will create the grammar does not need to add " " after each terminal and non-terminal.
10. Removal of code not related to grammar-based genetic programming (GGP).
11. GRCandidateProgram: Definition of two methods, named setFitness and getFitnessValue

If you are using this framework, please cite its paper:

=> F. Otero, T. Castle, and C. Johnson. EpochX: genetic programming in java with statistics and event monitoring. In Proceedings of the 14th annual conference companion on Genetic and evolutionary computation (GECCO Companion), ACM, New York, NY, USA, 93-100, 2012. DOI: https://doi.org/10.1145/2330784.2330800.