A Brief Overview of Swarm Intelligence-Based Algorithms for Numerical Association Rule Mining

Variable-Length Differential Evolution for Numerical and Discrete Association Rule Mining

“…In the context of stochastic population-based natureinspired algorithms, Evolutionary Algorithms (EAs), like Genetic Algorithms (GAs), and Swarm Intelligence (SI) based algorithms, like Particle Swarm Optimization (PSO), which were often utilized for rule mining tasks, have been adapted to mine numerical association rules efficiently [8], [9]. These algorithms encode potential solutions as chromosomes, allowing variation operators such as crossover and mutation to refine the rule population iteratively.…”

“…However, approaches are also based on the Ant Colony Optimization (ACO) [12], Bat algorithm (BA) [13], Cuckoo Search (CS) [14] and Differential Evolution (DE) [15], among others. Even though some algorithms work well on specific datasets, while others work on different datasets, much attention is nowadays given to developing new metrics that help obtain the most appropriate association rules [5], [9].…”

NarmViz: A novel method for visualization of time series numerical association rules for smart agriculture

“…Several cases of use and applications of ARM exist in the smart agriculture field (Fister & Salcedo‐Sanz, 2022; Fister Jr. et al, 2023; Khan & Singh, 2014; Rajesh, 2011). ARM methods are very efficient for providing new knowledge, represented as mathematical implications consisting of two parts, that is, an antecedent and a consequence (Fister Jr & Fister, 2020). Classical ARM methods can work only on datasets with discrete attributes, but modern methods are also tailored to work with mixed attributes in datasets, that is, numerical and discrete.…”

“…Classical ARM methods can work only on datasets with discrete attributes, but modern methods are also tailored to work with mixed attributes in datasets, that is, numerical and discrete. These methods are known under the name of numerical association rule mining (NARM) (Fister Jr & Fister, 2020). They are primarily based on stochastic population‐based nature‐inspired (NI) algorithms (Eiben & Smith, 2015), efficient methods for dealing with huge search spaces.…”

“…The TS‐NARM can be implemented as a NI algorithm (Fister Jr et al, 2013). Two classes belong to this kind of algorithm, that is, evolutionary (EAs) (Eiben & Smith, 2015) and SI based (Blum & Merkle, 2008).…”