Design and implementation of ANFIS on FPGA and verification with class classification problems

Abstract

In this research, we implemented ANFIS (Adaptive-Network-Based Fuzzy Inference System) on an FPGA as a System-on-Chip (SoC) system and evaluated its performance using iris flower classification. Additionally, we developed and tested an IEEE 754 16-bit floating-point format tailored for ANFIS operations. The evaluation focused on computational accuracy and speed. The results demonstrate that this FPGA-based ANFIS implementation, coupled with the 16-bit floating-point format, yields high efficiency and superior computational performance.