Gene expression data obtained from DNA microarrays are very useful in revealing the mechanisms that drive life. It is necessary to analyze these data through the use of algorithms, as in clustering and machine-learning. In a previous study, we developed fuzzy adaptive resonance theory (FuzzyART) and applied it to gene expression data, to identify genetic networks. FuzzyART was used as a clustering algorithm that is very suitable for the analysis of biological data; however, although FuzzyART is very useful in the analysis of dozens of gene expression profiles, it is difficult to apply this method to thousands of gene expression profiles, owing to inherent category proliferation and long calculation time. In the present study, we developed a knowledge-based FuzzyART (KB-FuzzyART) to mitigate these problems. We first constructed a gene list-1 from the gene database of Arabidopsis thaliana as knowledge for KB-FuzzyART, because KB-FuzzyART requires any knowledge as input. This method was applied to gene expression data obtained via the microarray analysis of A. thaliana, to identify the downstream genes of ASYMMETRIC LEAVES1 (AS1) and ASYMMETRIC LEAVES2 (AS2), both of which are involved in leaf development. The results of the analysis using KB-FuzzyART showed that the KNAT6 and YABBY5 (YAB5) genes are candidates for downstream factors, after a short calculation time for analysis. These results suggest that our gene list-1 is a very useful database for analyzing the expression profiles of genes that are related to the development of A. thaliana; they also suggest that the KB-FuzzyART has the high potential to function as a new method by which one can select candidate genes from thousands of genes, using gene expression data on mutant strains.