Yujun Li’s research while affiliated with Shandong University and other places

Sequential recommendation has attracted a lot of attention from both academia and industry, however the privacy risks associated with gathering and transferring users’ personal interaction data are often underestimated or ignored. Existing privacy-preserving studies are mainly applied to traditional collaborative filtering or matrix factorization rather than sequential recommendation. Moreover, these studies are mostly based on differential privacy or federated learning, which often lead to significant performance degradation, or have high requirements for communication. In this work, we address privacy-preserving from a different perspective. Unlike existing research, we capture collaborative signals of neighbor interaction sequences and directly inject indistinguishable items into the target sequence before the recommendation process begins, thereby increasing the perplexity of the target sequence. Even if the target interaction sequence is obtained by attackers, it is difficult to discern which ones are the actual user interaction records. To achieve this goal, we introduce a novel sequential recommender system called CLOUD ( C o L laborative-c O nfusion seq U ential recommen D er), which incorporates a collaborative confusion mechanism to modify the raw interaction sequences before conducting recommendation. Specifically, CLOUD first calculates the similarity between the target interaction sequence and other neighbor sequences to find similar sequences. Then, CLOUD considers the shared representation of the target sequence and similar sequences to determine the operation to be performed: keep, delete, or insert. A copy mechanism is designed to make items from similar sequences have a higher probability to be inserted into the target sequence. Finally, the modified sequence is used to train the recommender and predict the next item. We conduct extensive experiments on three benchmark datasets. The experimental results show that CLOUD achieves a maximum modification rate of 66.57% on interaction sequences, and obtains over 99% recommendation accuracy compared to the state-of-the-art sequential recommendation methods. This proves that CLOUD can effectively protect user privacy at minimal recommendation performance cost, which provides a new solution for privacy-preserving for sequential recommendation. Our implementation is available at https://github.com/weiwang0927/CLOUD .

The widespread deployment of dual-camera systems has laid a solid foundation for practical applications of infrared (IR)-RGB cross-modality person reidentification (ReID). However, the inherent modality differences between RGB and IR images cause significant intra-class variances in the feature space for individuals of the same identity. Current methods typically employ various network architectures for the image style transfer or extracting modality-invariant features, yet they overlook the information extraction from the most fundamental spectral semantic features. Based on the existing approaches, we propose a multi-spectral semantic alignment (MSSA) architecture aimed at aligning fine-grained spectral semantic features across both intra-modality and inter-modality perspectives. Through modality center semantic alignment (MCSA) learning, we comprehensively mitigate differences in identity features of different modalities. Moreover, to attenuate the discriminative information unique to a single modality, we introduce the modality reliability intensification (MRI) loss to enhance the reliability of identity information. Finally, to tackle the challenge that inter-modality intra-class disparities surpass inter-modality inter-class differences, we leverage the dynamic discriminative center (DDC) loss to further bolster the discriminability of reliable information. Through an extensive experiments conducted on SYSU-MM01, RegDB, and LLCM datasets, we demonstrate the substantial advantages of the proposed MSSA over other state-of-the-art methods.

Device-to-device (D2D) communication enables direct data transmissions between devices, thereby filling in the blind zones of base stations (BSs) and alleviating their load pressures. In this work, we propose an adaptive single-hop and relay-selection-based double-hop D2D transmission scheme underlaying cellular networks with full-frequency reuse. When the communication distance between two D2D users is shorter than a threshold $d_{0}$ , the single-hop D2D link can be established. However, when the distance between two D2D users is longer than $d_{0}$ but shorter than $2d_{0}$ , and there exists at least one relay node in the overlapping coverage areas of both users, the double-hop D2D transmission is performed. Based on the stochastic geometry theory, we properly model the random distributions of D2D users, relay nodes, cellular users, and BSs, as well as the aggregate interference for each type of communication mode. We propose to divide the overlapping area of the coverage regions of two D2D users into small grids, and the relay selection priority is assigned to each grid according to the selection criterion. The griding of overlapping areas can greatly simplify the analysis of successful relaying by considering the possible locations of the selected relay. We also present two benchmark schemes, namely cellular only, cellular with D2D single-hop. Numerical results show that our proposed scheme can significantly improve the area throughput compared with the benchmark schemes.

generation technology can assist administrative law enforcement officers in grasping the core summary of administrative cases more quickly, which is one of the research focuses in the applied research of administrative justice. How to cope with the phenomenon of long-distance dependence on judicial documents and realize high-precision summary generation is an urgent problem to be solved at present. We propose a segmented judicial instrument summary generation model to solve the long-distance dependency problem. The model adopts a two-segment structure combining an extractive summary model and a generative summary model. Among them, the extractive summarization model consists of two parts: the BERT model and the document-level sentence encoder of the Transformer structure, which is used to initially process the judicial documents, extract the key legal information in the original documents, and compose the preliminary summaries, which are used as the training corpus for the generative summarization model. In addition, we introduce the contrast learning idea to alleviate the exposure bias problem in the generative model. The generative summary model takes T5-PEGASUS as the main body, generates pseudo-optimal summaries and multiple candidate summaries, and constructs contrast loss based on contrast learning to mitigate the exposure bias problem in the main model. Finally, the effectiveness of the algorithm is verified through experiments on the CAIL 2020 dataset (Challenge of AI in Law 2020) and the self-labelled AED summary dataset dedicated to administrative enforcement documents (Administrative Enforcement Documents Dataset).