Knowledge graphs (KGs), with their structured representation capabilities, offer promising avenue for enhancing Retrieval Augmented Generation (RAG) systems, leading to the development of KG-RAG systems. Nevertheless, existing methods often struggle to achieve effective synergy between system effectiveness and cost efficiency, leading to neither unsatisfying performance nor excessive LLM prompt tokens and inference time. To this end, this paper proposes REMINDRAG, which employs an LLM-guided graph traversal featuring node exploration, node exploitation, and, most notably, memory replay, to improve both system effectiveness and cost efficiency. Specifically, REMINDRAG memorizes traversal experience within KG edge embeddings, mirroring the way LLMs "memorize" world knowledge within their parameters, but in a train-free manner. We theoretically and experimentally confirm the effectiveness of REMINDRAG, demonstrating its superiority over existing baselines across various benchmark datasets and LLM backbones. Our code is available at https://github.com/kilgrims/ReMindRAG.

This paper proposes a novel multi-cluster network system for autonomous underwater vehicle (AUV) swarm communication and positioning, addressing challenges of low throughput, high end-to-end delay, and significant ranging errors in traditional asynchronous underwater acoustic networks. The proposed system introduces a novel hybrid Medium Access Control (MAC) protocol, which combines intra-cluster Time Division Multiple Access (TDMA) with inter-cluster Frequency Division Multiple Access (FDMA) to enhance channel utilization. Additionally, the system introduces an adaptive time-offset estimation method based on network data exchange and the Ultra-Short Baseline Moving Ranging (USBL-MR) algorithm, which significantly improves network throughput and reduces delay and improves positioning accuracy. Field trials demonstrate that the system achieves a throughput of 398.3 bps with an end-to-end delay of 21.2 s. Furthermore, the ranging Root Mean Square Error (RMSE) has been reduced from 7.87 to 2.42 m, marking a 14% improvement in positioning accuracy.

One major goal of the AI security community is to securely and reliably produce and deploy deep learning models for real-world applications. To this end, data poisoning based backdoor attacks on deep neural networks (DNNs) in the production stage (or training stage) and corresponding defenses are extensively explored in recent years. Ironically, backdoor attacks in the deployment stage, which can often happen in unprofessional users' devices and are thus arguably far more threatening in real-world scenarios, draw much less attention of the community. We attribute this imbalance of vigilance to the weak practicality of existing deployment-stage backdoor attack algorithms and the insufficiency of real-world attack demonstrations. To fill the blank, in this work, we study the realistic threat of deployment-stage backdoor attacks on DNNs. We base our study on a commonly used deployment-stage attack paradigm - adversarial weight attack, where adversaries selectively modify model weights to embed backdoor into deployed DNNs. To approach realistic practicality, we propose the first gray-box and physically realizable weights attack algorithm for backdoor injection, namely subnet replacement attack (SRA), which only requires architecture information of the victim model and can support physical triggers in the real world. Extensive experimental simulations and system-level real- world attack demonstrations are conducted. Our results not only suggest the effectiveness and practicality of the proposed attack algorithm, but also reveal the practical risk of a novel type of computer virus that may widely spread and stealthily inject backdoor into DNN models in user devices. By our study, we call for more attention to the vulnerability of DNNs in the deployment stage.

Operation safety and efficiency are two main concerns in power plants. It is important to detect the anomalies in power plants, and further judge whether they affect key performance indicators (KPIs), such as the thermal efficiency. These two goals can be achieved by KPI-related nonstationary process monitoring. Although the thermal efficiency cannot be accurately measured online, it can be strongly characterized by some online measurable variables, including the exhaust gas temperature and oxygen content of flue gas. These critical variables closely related to the thermal efficiency are termed as output variables. Inspired from nonstationary common trends between input and output variables in thermal power plants, the output-relevant common trend analysis (OCTA) method is proposed, in this article, to model the input–output relationship. In OCTA, input and output variables are decomposed into nonstationary common trends and stationary residuals, and the model parameters are estimated by solving an optimization problem. It is pointed out that OCTA is a generalized form of partial least squares (PLS). The superior monitoring performance of OCTA is illustrated by case studies on a real power plant in Zhejiang Provincial Energy Group of China. Compared with the other PLS-based recursive algorithms, OCTA can effectively detect the anomalies in power plants and accurately determine whether they have an impact on the thermal efficiency or not.

In the die bonding process of MEMS microphone chip, the exact center and rotate angle of MEMS microphone chip is corrected after the vacuum nozzle has adsorbed it. In order to achieve high speed and precision placement function, the flying vision system based on rotating mirror is proposed, the mirror is driven to rotate by the power drawn from the nozzle rise and fall movement through a rack and pinion mechanism, the nozzle pick and place functions is not interfered by mirror through designing the rack and pinion mechanism parameters. The vision system can capture an static clear MEMS chip image when the mirror rotated to a 45 degrees position referenced with horizontal position, the center position and rotating angle of MEMS chip is fast calculated by the fine and coarse Hough Transform Algorithm, the center location and angle will be corrected by the control system during the mounted head is flying to placement position. The experimental results show that the system can obtain high-quality MEMS microphone chip image, the chip's center position and rotating angle correction algorithm takes no more than 0.1s, so it can meet the high-speed placement requirements. In this proposed flying vision system, mechanical structure is simple and compact, MEMS chip's image is clear without distortion, and there is no restriction on the mounted head movement trajectory. The flying vision system significantly enhances the quality and efficiency of SMD Chip; it can be widely used in the surface mounted system.