Heuristic algorithms for obtaining Polynomial Threshold Functions with low densities
This work addresses the challenge of optimizing PTF representations for Boolean functions, which is incremental as it builds on prior algorithms like Oztop's.
The paper tackled the problem of finding polynomial threshold function (PTF) representations of Boolean functions with minimal monomials, and the result showed that their heuristic algorithms, including a Genetic Algorithm, produced more parsimonious representations compared to existing non-heuristic and GA-based methods.
In this paper we present several heuristic algorithms, including a Genetic Algorithm (GA), for obtaining polynomial threshold function (PTF) representations of Boolean functions (BFs) with small number of monomials. We compare these among each other and against the algorithm of Oztop via computational experiments. The results indicate that our heuristic algorithms find more parsimonious representations compared to the those of non-heuristic and GA-based algorithms.