Recently, the research team headed by Associate Professor Wang Zhongrui from the School of Microelectronics at the Southern University of Science and Technology, working alongside Researcher Shang Dashan and Researcher Xu Xiaoxin from Academician Liu Ming's team at the Institute of Microelectronics, Chinese Academy of Sciences, has achieved remarkable advancements in the realm of analog artificial intelligence-optimized resistive memory. The team has introduced a software-hardware co-design framework that leverages edge-pruning topological optimization. This innovative approach effectively tackles key challenges, including the randomness inherent in device programming, non-linearity, as well as the high energy consumption and time costs associated with analog computing in randomly weighted resistive memory neural networks. Drawing inspiration from the brain's Hebbian learning principle (Hebb's rule) and its structural plasticity, this method refines network topology by 'preserving effective connections and eliminating redundant edges.' It capitalizes on the natural randomness of the electroforming process in resistive memory to generate extensive random weights at a minimal cost.