Lijuan Zhang

In this letter, we propose a novel structure-mapped prototype clustering (SMPC) method for blind detection in reconfigurable intelligent surface (RIS)-assisted spatial modulation (RIS-SM) systems. In RIS-SM scenarios, acquiring accurate channel state information (CSI) is often impractical due to the passive nature of RIS, which motivates the use of unsupervised clustering techniques for blind signal detection. However, existing clustering-based detectors often suffer from redundant parameter estimation and strong dependence on a large number of received signals. In contrast, SMPC reformulates the clustering task from a structure-driven perspective by modeling the signal constellation as a geometric template defined by a set of transformation mappings. A complete set of cluster centers is reconstructed from a minimal number of prototype centroids using these transformations. Moreover, by mapping received signals into a unified prototype space via inverse transformations, SMPC improves clustering compactness and remains robust even when only a small number of received signals are available. The proposed framework generalizes well to arbitrary modulation schemes with geometric symmetry, enabling consistent and efficient detection without requiring CSI. Simulation results demonstrate that SMPC achieves near-optimal bit error rate (BER) performance and maintains robust detection capability under unknown CSI conditions.

In this letter, we propose a novel blind detector for RIS-assisted received spatial modulation (RIS-RSM) systems utilizing a clustering-based approach. The combination of reconfigurable intelligent surfaces (RIS) and spatial modulation (SM) presents a promising direction for beyond 5G (B5G) networks, enhancing spectral and energy efficiency. However, existing signal detection methods for RIS-RSM assume perfect channel state information (CSI), which is impractical due to the passive nature of RIS. To overcome this, we first transform the blind detection problem of RIS-RSM into a clustering problem and apply unsupervised clustering algorithms from machine learning to eliminate the need for CSI acquisition. Given that traditional clustering algorithms like K-means are insufficient for this application, we propose a novel clustering detector by leveraging the unique amplitude and phase characteristics of the channel in RIS-RSM systems. Simulation results demonstrate that our proposed detector can maintain excellent detection performance that is basically consistent with the optimal detector ML without the need for CSI, marking a significant advancement in RIS-RSM signal detection.

Recently, Arıkan proposed a polarization‐adjusted convolutional (PAC) codes, demonstrating their superior error correction performance over polar codes at short block lengths. It was confirmed that PAC codes approached the optimal performance achievable with limited code length. This paper proposes a novel low‐complexity list decoding algorithm for PAC codes, incorporating path splitting and pruning strategies based on a set of highly reliable information bits. Simulation results reveal that the proposed algorithm significantly reduces sorting complexity and average list size, all while incurring negligible performance loss. Unlike previous pruning algorithms designed for polar codes, the proposed strategy eliminates the need to individually assess the reliability of decoding paths in each decoding process. Instead, the algorithm minimizes redundant decoding paths through a high‐reliability information bit set, constructed using Monte Carlo experiments.

The task of named entity recognition has been widely used in medical text analysis, but there is still the problem of poor transfer ability in practical applications. This work proposes a novel named entity recognition method based on a proposed knowledge distillation framework and Efficient GlobalPointer for Chinese biomedical and clinical data. Specifically, our study leverages the Efficienct GlobalPointer to address the issue of entity nesting and introduces a context shield window to mitigate interference from redundant information. Furthermore, the model’s generalization ability is enhanced through a novel knowledge distillation framework. The proposed knowledge distillation framework solves the problem of independent feature learning process in feature distillation by using linkage mechanism. The recognition accuracy is improved by the proposed knowledge distillation method while keeping the model complexity low, so that our method can meet the inference speed requirement in real applications while ensuring a certain recognition accuracy. Our method achieves excellent experimental results on three publicly available Chinese datasets, where the comprehensive evaluation metric F1 exceeds the best results achieved by existing methods.

Medication recommendation based on electronic health records (EHRs) is a significant research direction in the biomedical field, which aims to provide a reasonable prescription for patients according to their historical and current health conditions. However, the existing recommended methods have many limitations in dealing with the structural and temporal characteristics of EHRs. These methods either only consider the current state while ignoring the historical situation, or fail to adequately assess the structural correlations among various medical events. These factors result in poor recommendation quality. To solve this problem, we propose an augmented graph structural–temporal convolutional network (A-GSTCN). Firstly, an augmented graph attention network is used to model the structural features among medical events of patients’ EHRs. Next, the dilated convolution combined with residual connection is applied in the proposed model, which can improve the temporal prediction capability and further reduce the complexity. Moreover, the cache memory module further enhances the model’s learning of the history of EHRs. Finally, the A-GSTCN model is compared with the baselines through experiments, and the efficiency of the A-GSTCN model is verified by Jaccard, F1 and PRAUC. Not only that, the proposed model also reduces the training parameters by an order of magnitude.