GARA: A novel approach to Improve Genetic Algorithms' Accuracy and Efficiency by Utilizing Relationships among Genes

TL;DR

This work addresses the limitation of traditional genetic algorithms that treat genes independently by introducing GRGA, which models inter-gene relationships with a directed RGGR to guide crossover and mutation. The RGGR is updated adaptively based on fitness, promoting favorable gene interactions via a first-order Markov process and influencing locus selection with formulas such as $s_{(i,j)}^{(k,k+1)} = \frac{1}{C_1 + C_2 W_{(i,j)}^{(k,k+1)}}$ and $P_{crossover} = \frac{S_{i,j_1j_2}^{(k,k+1)}}{\sum_k S_{i,j_1j_2}^{(k,k+1)}} \times 100\%$, $P_{mutation} = \frac{s_{i,j}^{(k,k+1)}}{\sum_k s_{i,j}^{(k,k+1)}} \times 100\%$. The method is validated on the Shubert function and three applications—feature selection, text summarization, and dimensionality reduction—demonstrating faster convergence and competitive performance, with notable efficiency gains and improved quality in several tasks. The results indicate that RGGR-guided loci can significantly enhance both exploration and exploitation, and GRGA offers a general framework for extending existing GAs. Future work includes extending RGGR to higher-order Markov relationships to capture more complex gene interactions.

Abstract

Genetic algorithms have played an important role in engineering optimization. Traditional GAs treat each gene separately. However, biophysical studies of gene regulatory networks revealed direct associations between different genes. It inspires us to propose an improvement to GA in this paper, Gene Regulatory Genetic Algorithm (GRGA), which, to our best knowledge, is the first time to utilize relationships among genes for improving GA's accuracy and efficiency. We design a directed multipartite graph encapsulating the solution space, called RGGR, where each node corresponds to a gene in the solution and the edge represents the relationship between adjacent nodes. The edge's weight reflects the relationship degree and is updated based on the idea that the edges' weights in a complete chain as candidate solution with acceptable or unacceptable performance should be strengthened or reduced, respectively. The obtained RGGR is then employed to determine appropriate loci of crossover and mutation operators, thereby directing the evolutionary process toward faster and better convergence. We analyze and validate our proposed GRGA approach in a single-objective multimodal optimization problem, and further test it on three types of applications, including feature selection, text summarization, and dimensionality reduction. Results illustrate that our GARA is effective and promising.

Figures (8)

• Figure 1: Interactions between genes. a. The expression product of the LEAFY (LFY) gene regulates APETALA1 (AP1) expression. b.The sequential process from tyrosine input to the final production of adrenaline.
• Figure 2: The working principle of our GRGA. a. Relationship graph representing gene regulation (RGGR), where the weight value in the edge represents the relationship degree between each part of the solution is updated from the fitnesses of individuals. b. Crossover loci were selected by RGGR. c. Mutation locus was selected by RGGR.
• Figure 3: We give an example showing that when two gene loci are shared by two individuals, this is how the edges connected by these two individuals calculate the value of W.
• Figure 4: How to choose crossover and mutation loci. We provide an example to illustrate the crossover and mutation processes of GRGA in detail.
• Figure 5: The evolution of maximum and average fitness of GRGA and original GA over 100 experiments with evolving generations.