The paper "Rewrite the Stars" by Xu Ma, Xiyang Dai, Yue Bai, Yizhou Wang, and Yun Fu explores the underutilized potential of the "star operation" (element-wise multiplication) in network design. The authors argue that this operation can map inputs into high-dimensional, non-linear feature spaces without increasing the network's width, similar to kernel tricks. They introduce StarNet, a simple yet powerful prototype that demonstrates impressive performance and low latency under compact network structures and efficient budgets. The study reveals that the star operation can generate a new feature space with approximately $(\frac{d}{\sqrt{2}})^2$ linearly independent dimensions, where $d$ is the input channel number. This capability allows StarNet to outperform various efficient models like MobileNetv3, EdgeViT, and FasterNet. The paper also discusses the implications of the star operation for efficient network designs and highlights its potential in areas such as activation-free networks and self-attention. The authors provide empirical, theoretical, and visual evidence to support their findings and encourage further exploration of the star operation in various tasks.The paper "Rewrite the Stars" by Xu Ma, Xiyang Dai, Yue Bai, Yizhou Wang, and Yun Fu explores the underutilized potential of the "star operation" (element-wise multiplication) in network design. The authors argue that this operation can map inputs into high-dimensional, non-linear feature spaces without increasing the network's width, similar to kernel tricks. They introduce StarNet, a simple yet powerful prototype that demonstrates impressive performance and low latency under compact network structures and efficient budgets. The study reveals that the star operation can generate a new feature space with approximately $(\frac{d}{\sqrt{2}})^2$ linearly independent dimensions, where $d$ is the input channel number. This capability allows StarNet to outperform various efficient models like MobileNetv3, EdgeViT, and FasterNet. The paper also discusses the implications of the star operation for efficient network designs and highlights its potential in areas such as activation-free networks and self-attention. The authors provide empirical, theoretical, and visual evidence to support their findings and encourage further exploration of the star operation in various tasks.