The paper "On-Device Recommender Systems: A Comprehensive Survey" by Hongzhi Yin, Liang Qu, Tong Chen, Wei Yuan, Ruiqi Zheng, Jing Long, Xin Xia, Yuhui Shi, and Chengqi Zhang provides a comprehensive overview of on-device recommender systems (DeviceRSs). The authors address the limitations of traditional cloud-based recommender systems (CloudRSs), which suffer from high resource consumption, response latency, and privacy and security risks. DeviceRSs leverage edge devices to minimize centralized data storage, reduce communication overheads, and enhance user privacy and security. The survey covers three main aspects: 1. **Deployment and Inference**: Techniques for compressing large recommendation models to fit resource-constrained devices, including binary code-based methods, embedding sparsification, variable size embeddings, compositional embeddings, and sustainable deployment methods. 2. **Training and Update**: Methods for local data-driven model optimization, such as federated learning, P2P collaborative training, and fine-tuning pre-trained models using local data. 3. **Security and Privacy**: Strategies to protect user data and models from malicious attacks, including privacy-preserving techniques and defense mechanisms. The authors propose a systematic taxonomy of DeviceRS methods and discuss the challenges and future research directions in the field. This survey aims to bridge the gap in literature reviews and provide a comprehensive understanding of the current state of DeviceRSs, helping researchers and practitioners effectively grasp the latest advancements and develop new ideas for this emerging technology.The paper "On-Device Recommender Systems: A Comprehensive Survey" by Hongzhi Yin, Liang Qu, Tong Chen, Wei Yuan, Ruiqi Zheng, Jing Long, Xin Xia, Yuhui Shi, and Chengqi Zhang provides a comprehensive overview of on-device recommender systems (DeviceRSs). The authors address the limitations of traditional cloud-based recommender systems (CloudRSs), which suffer from high resource consumption, response latency, and privacy and security risks. DeviceRSs leverage edge devices to minimize centralized data storage, reduce communication overheads, and enhance user privacy and security. The survey covers three main aspects: 1. **Deployment and Inference**: Techniques for compressing large recommendation models to fit resource-constrained devices, including binary code-based methods, embedding sparsification, variable size embeddings, compositional embeddings, and sustainable deployment methods. 2. **Training and Update**: Methods for local data-driven model optimization, such as federated learning, P2P collaborative training, and fine-tuning pre-trained models using local data. 3. **Security and Privacy**: Strategies to protect user data and models from malicious attacks, including privacy-preserving techniques and defense mechanisms. The authors propose a systematic taxonomy of DeviceRS methods and discuss the challenges and future research directions in the field. This survey aims to bridge the gap in literature reviews and provide a comprehensive understanding of the current state of DeviceRSs, helping researchers and practitioners effectively grasp the latest advancements and develop new ideas for this emerging technology.