With the widespread application of large language models (LLMs) in natural language processing (NLP), hallucinations have become a significant impediment to their effective use of LLMs in industry applications. To address this challenge, we integrate existing hallucination detection and mitigation methods into a unified hallucination detection and mitigation framework. The framework consists of four main components: output parser, reference parser, fact verifier, and mitigator. These components collectively consolidate various hallucination detection and mitigation methods. Within this unified framework, we introduce the hierarchical semantic piece (HSP) for hallucination detection and mitigation. The HSP method extracts multi-granularity semantic pieces from both the reference material and the generated text. Sentence-level semantic pieces encapsulate global semantic information, while entity-level semantic pieces handle local semantic information. This method verifies the consistency between the generated text and the reference text at corresponding granularities, thereby enhancing the effectiveness of hallucination detection and mitigation. Experimental results show that the HSP method is very effective in detecting and mitigating hallucinations and shows lower computational resource consumption. Our method has great potential and promises for industry applications that rely on professionalism and reliability.
A physical layer security scheme based on cooperative relays is proposed to address the problem of eavesdropping in satellite downlink which is induced by the broadcasting nature of the shared communication medium and the long distance of the satellite-ground transmissions. By evaluating the outage probability, the relay nodes participating in cooperative communication can be divided into a cooperative forwarding group and a cooperative jamming group. The cooperative jamming group and the cooperative forwarding groups construct a cascaded channel which combined with the random weighting technique improves the quality of transmission on the main channel. The cooperative jamming group uses the main channel null space to transmit artificial noise which does not affect the quality of the main channel, but reduces the eavesdropping channel quality. Simulation results show that the proposed anti-eavesdropping scheme can maximize the quality difference between the main channel and the eavesdropping channels, which thereby may significantly improve the security capacity of satellite downlink systems.
14 January 2021 in IEEE Geoscience and Remote Sensing Letters
Object detection in aerial images is a challenging task because of the complex background and various orientations of the objects. Currently, many detectors have made significant progress in improving mAP scores, but they do not achieve the improvements in efficiency and model size. In this letter, we propose a detector for resource-limited satellite network, namely, the simple convolutional neural networks (simple-CNNs), which can be directly applied in actual application scenarios using small sample data. Specifically, a new sample 16-layer network is devised to reduce the model size. Meanwhile, we calculate a new anchor for ten classes and carefully design a change-IOU Loss (CI-Loss) function for horizontal bounding box detection to improve the accuracy. Extensive experiments on two remote sensing public data sets (NWPU VHR-10 and part of DIOR) show good performance for the efficiency of our detector. In particular, our simple-CNN achieves 68.9 mAP on the NWPU VHR-10 test-dev data set with 90 M parameters and a test speed of 72 ms, and it also has good performance on a part of DIOR.
Satellite edge computing (SEC) has become a revolutionary paradigm to improve the quality of service, reduce the pressure on satellite-terrestrial backhaul bandwidth and reduce the average response delay of task requests. In this paper, we propose a task offloading algorithm based on K-D3QN to meet the rapidly growing demand of ground users. This algorithm improves the DQN algorithm by incorporating a satellite resource clustering module, a DDQN algorithm, and a competitive network mechanism module. The offloading decision-making process comprehensively considers three optimization objectives: task latency, resource utilization, and load-balancing degree, to achieve dynamic multi-objective optimization. Experimental results shown that the algorithm significantly reduces task latency, improves resource utilization and load-balancing degree.
In Space–Air–Ground Integrated Networks (SAGIN), computation offloading technology is a new way to improve the processing efficiency of node tasks and improve the limitation of computing storage resources. To solve the problem of large delay and energy consumption cost of task computation offloading, which caused by the complex and variable network offloading environment and a large amount of offloading tasks, a computation offloading decision scheme based on Markov and Deep Q Networks (DQN) is proposed. First, we select the optimal offloading network based on the characteristics of the movement of the task offloading process in the network. Then, the task offloading process is transformed into a Markov state transition process to build a model of the computational offloading decision process. Finally, the delay and energy consumption weights are introduced into the DQN algorithm to update the computation offloading decision process, and the optimal offloading decision under the low cost is achieved according to the task attributes. The simulation results show that compared with the traditional Lyapunov-based offloading decision scheme and the classical Q-learning algorithm, the delay and energy consumption are respectively reduced by 68.33% and 11.21%, under equal weights when the offloading task volume exceeds 500 Mbit. Moreover, compared with offloading to edge nodes or backbone nodes of the network alone, the proposed mixed offloading model can satisfy more than 100 task requests with low energy consumption and low delay. It can be seen that the computation offloading decision proposed in this paper can effectively reduce the delay and energy consumption during the task computation offloading in the Space–Air–Ground Integrated Network environment, and can select the optimal offloading sites to execute the tasks according to the characteristics of the task itself.
Syntheticaperture radar (SAR) object detection offers significant advantages in remote sensing applications, particularly under adverse weather conditions or low-light environments. However, single-modal SAR image object detection encounters numerous challenges, including speckle noise, limited texture information, and interference from complex backgrounds. To address these issues, we present Modality-Aware Adaptive Interaction Enhancement Network (MAIENet), a multimodal detection framework designed to effectively extract complementary information from both SAR and optical images, thereby enhancing object detection performance. MAIENet comprises three primary components: batch-wise splitting and channel-wise concatenation (BSCC) module, modality-aware adaptive interaction enhancement (MAIE) module, and multi-directional focus (MF) module. The BSCC module extracts and reorganizes features from each modality to preserve their distinct characteristics. The MAIE module component facilitates deeper cross-modal fusion through channel reweighting, deformable convolutions, atrous convolution, and attention mechanisms, enabling the network to emphasize critical modal information while reducing interference. By integrating features from various spatial directions, the MF module expands the receptive field, allowing the model to adapt more effectively to complex scenes. The MAIENet framework is end-to-end trainable and can be seamlessly integrated into existing detection networks with minimal modifications. Experimental results on the publicly available OGSOD-1.0 dataset demonstrate that MAIENet achieves superior performance compared with existing methods, achieving 90.8% .