Transformer-empowered Graph Neural Networks (TF-GNNs) are gaining significant attention in AI research because they leverage the front-end Transformer's ability to process textual data while also harnessing the back-end GNN's capacity to analyze graph structures.Typically, TF-GNNs follow the sequential execution mode, where the front-end Transformer first encodes vertex features, followed by subgraph sampling and subsequent processing by the back-end GNN.However, due to the massive computation workloads of Transformers and the irregular memory access patterns of GNNs, achieving efficient inference for TF-GNNs remains a challenge.Although architectures like FACT and MEGA have been proposed to separately accelerate the Transformer and GNN, they overlook the new opportunities arising from the coupling of the Transformer and GNN.To enable efficient TF-GNNs, we propose HEAT, a heterogeneous architecture with a Neural Processing Unit (NPU) and a DIMM-based Near-Data Processing (NDP).Such a heterogeneous architecture can utilize both the high computational power of NPU and the high internal bandwidth of NDP.To fully unleash the potential of the NPU-NDP architecture, HEAT makes the following three contributions: First, HEAT leverages graph topology to identify the importance of vertices and encodes their features in the Transformer using varying precision accordingly.Second, HEAT gives more flexibility to the execution granularity and execution order of * Zhuoran Song is the corresponding author.

Radio frequency (RF) fingerprint recognition uses the hardware characteristic differences of wireless devices to achieve identity authentication from the physical layer. However, there are significant differences in the distribution of the RF fingerprint feature of the same transmitter signal collected in different receiving devices. The existing deep learning algorithms usually retrain the model to improve the recognition accuracy, which brings challenges such as the high cost of retraining and a lack of training samples. Transfer learning can promote the learning of new tasks by extracting knowledge from existing tasks, and can be used to reduce feature distribution differences between different data sets. In this paper, we optimize the CycleGAN network, which is widely used in transfer learning, and propose a Dual Transfer-Generative Adversarial Network (DTGAN). This network replaces the transposed convolution with one-dimensional linear interpolation, which better aligns the feature boundaries of the target domain signal with the source domain signal. The experimental results show that without retraining the model, the recognition accuracy of data after feature transfer is significantly better than that of the non-transfer scenario, which enhances the robustness of RF fingerprint recognition under different receiving conditions.

Recently, there has been a surge in machine-generated natural language content being misused by unauthorized parties. Watermarking is a well-recognized technique to address the issue by tracing the provenance of the text. However, we found that most existing watermarking systems for texts are subject to ad hoc design and thus suffer from fundamental vulnerabilities.

To solve the increasingly prominent contradiction between the traditional damage control and the demand of high efficiency and reliability of ship system, a ship damage control system based on spatio-temporal database is presented and accomplished with cloud solution. A path planning algorithm based on Dijkstra is proposed to meet the dynamic road network as in the fire rescue scenario. The binary group is adopted to describe the weight of the path, and path network is pruning to reduce nodes accessed in advance. The simulation results show that the proposed algorithm improves the efficiency, capacity, intelligence and user experience, and provide efficient support for assistant decision-making.