Heuristic Algorithm Used by RES2DINV

RES2DINV, an evolutionary optimization algorithm, employs a heuristic approach to identify promising areas in the search space. This method leverages the concept of self-organization maps (SOMs) as an extracting knowledge technique to reduce the search space and improve convergence properties.

The algorithm begins by defining the population size, which corresponds to the Sphere function in this case. The fitness function used is SOM, the main function for the EA-SOM algorithm. This implementation was first introduced in a publication by Galvez et al., where it showed significant improvements in converging to high multi-modal optima using a reduced number of function evaluations.

The algorithm works by producing an entire population of candidate solutions using local information from each neural unit within the SOM lattice. Once trained, the best solution is identified, and a subset of generated solutions are selected to train the SOM in each generation. This approach enables the algorithm to extract useful knowledge during its operation, guiding its search strategy toward promising regions.

In contrast to traditional evolutionary techniques that produce new solutions without extracting useful knowledge from previously generated solutions, this method utilizes acquired knowledge to improve performance and convergence properties. Evolutionary approaches have been widely applied to complex optimization problems in various domains, offering an effective alternative solution.

One of the key advantages of SOMs is their ability to map high-dimensional relations onto a reduced lattice while preserving the topological organization of the initial data. This property makes them particularly useful for extracting knowledge from large datasets.

In summary, RES2DINV's heuristic algorithm uses a SOM-based approach to extract knowledge and guide its search strategy toward promising regions. By leveraging this technique, the algorithm demonstrates improved convergence properties and reduced function evaluations, making it an attractive solution for complex optimization problems.

Heuristic algorithm used by res2dinv