Rewrite the Stars

2024/06/16 by Xu Ma, Xiyang Dai, Yue Bai +2 · 14 citations
Arts and Humanities · Computer Science · Physics and Astronomy · #Historical and Architectural Studies #Currency Recognition and Detection #Historical Astronomy and Related Studies

paper · doi:10.1109/cvpr52733.2024.00544

openalex publication_date 2024/06/16 · openalex created_date 2024/09/23 · openalex updated_date 2026/07/31

Abstract

Recent studies have drawn attention to the untapped potential of the “star operation” (element-wise multiplication) in network design. While intuitive explanations abound, the foundational rationale behind its application remains largely unexplored. Our study attempts to reveal the star operation's ability of mapping inputs into high-dimensional, non-linear feature spaces-akin to kernel tricks-without widening the network. We further introduce StarNet, a simple yet pow-erful prototype, demonstrating impressive performance and low latency under compact network structure and efficient budget. Like stars in the sky, the star operation appears unremarkable but holds a vast universe of potential. Our work encourages further exploration across tasks, with codes available at https://github.com/ma-xu/Rewrite-the-Stars.

Cited by

• Situat3DChange: Situated 3D Change Understanding Dataset for Multimodal Large Language Model
• Designing Practical Models for Isolated Word Visual Speech Recognition
• DeepTraverse: A Depth-First Search Inspired Network for Algorithmic Visual Understanding
• RadHARSimulator V1: Model-Based FMCW Radar Human Activity Recognition Simulator
• A Lightweight Group Multiscale Bidirectional Interactive Network for Real-Time Steel Surface Defect Detection
• E-ConvNeXt: A Lightweight and Efficient ConvNeXt Variant with Cross-Stage Partial Connections
• UniConvNet: Expanding Effective Receptive Field while Maintaining Asymptotically Gaussian Distribution for ConvNets of Any Scale
• Rate-distortion Optimized Point Cloud Preprocessing for Geometry-based Point Cloud Compression
• Exploration of Low-Cost but Accurate Radar-Based Human Motion Direction Determination
• LMEFA-YOLO: Remote sensing image object detection network based on lightweight feature enhancement and efficient fusion
• StarIO: A Lightweight Inertial Odometry for Nonlinear Motion
• FTCFormer: Fuzzy Token Clustering Transformer for Image Classification
• MobileIE: An Extremely Lightweight and Effective ConvNet for Real-Time Image Enhancement on Mobile Devices
• Efficient Star Distillation Attention Network for Lightweight Image Super-Resolution

Related

• Rewrite the Stars
• STAR: Synthesis of Tailored Architectures
• ResNetX: a more disordered and deeper network architecture
• Model Rubik's Cube: Twisting Resolution, Depth and Width for TinyNets
• Coarsening the Granularity: Towards Structurally Sparse Lottery Tickets