ABSTRACT Reconfigurable intelligent surface (RIS)‐assisted received spatial modulation (RIS‐RSM) has emerged as a promising technique to enhance spectral and energy efficiency in next‐generation wireless systems. However, performing accurate signal detection without channel state information (CSI) remains a critical challenge, particularly in blind detection scenarios. In this paper, we propose a novel clustering‐based blind detector named energy‐tiered structure initialization (ETSI). The proposed method exploits the amplitude heterogeneity of modulation symbols—originating from the unequal energy levels of QAM or hybrid constellations—by partitioning the signal space into multiple energy tiers. Each receive antenna is associated with a structure prototype, representing a normalized statistical channel pattern, which is initialized using the distribution of the underlying fading model (e.g., Rayleigh) rather than instantaneous CSI. During clustering, these prototypes are iteratively refined through tier‐wise averaging of normalized signal samples, thereby enforcing structural consistency and mitigating amplitude‐induced bias. After convergence, the constellation‐aligned cluster centres are reconstructed by combining the updated prototypes with their corresponding modulation amplitudes, inherently enabling the joint detection of both the modulation symbol and the RIS‐assisted spatial index. Simulation results show that ETSI achieves around 0.5–1 dB SNR gain over amplitude–phase aware clustering under 8PSK, and about 1–1.5 dB improvement under 16QAM, while outperforming the CSI‐based greedy detector (GD) across both modulations. Moreover, ETSI achieves BER performance close to that of the perfect‐CSI maximum likelihood detector, confirming its accuracy, scalability and practical feasibility for blind RIS‐RSM detection.

Hierarchical text classification (HTC) is an important task in natural language processing (NLP). Existing methods typically utilize both text features and the hierarchical structure of labels to categorize text effectively. However, these approaches often struggle with fine-grained labels, which are closely similar, leading to difficulties in accurate classification. At the same time, contrastive learning has significant advantages in strengthening fine-grained label features and discrimination. However, the performance of contrastive learning strongly depends on the construction of negative samples. In this paper, we design a hierarchical sequence ranking (HiSR) method for generating diverse negative samples. These samples maximize the effectiveness of contrastive learning to enhance the ability of the model to distinguish between fine-grained labels and improve the performance of the model in HTC. Specifically, we transform the entire label set into linear sequences based on the hierarchical structure and rank these sequences according to their quality. During model training, the most suitable negative samples were dynamically selected from the ranked sequences. Then contrastive learning amplifies the differences between similar fine-grained labels by emphasizing the distinction between the ground truth and the generated negative samples, thereby enhancing the discriminative ability of the model. Our method has been tested on three public datasets and achieves state-of-art (SOTA) on two of them, demonstrating its effectiveness.

Transformer-based approaches have gained significant attention in image restoration, where the core component, i.e, Multi-Head Attention (MHA), plays a crucial role in capturing diverse features and recovering high-quality results. In MHA, heads perform attention calculation independently from uniform split subspaces, and a redundancy issue is triggered to hinder the model from achieving satisfactory outputs. In this paper, we propose to improve MHA by exploring diverse learners and introducing various interactions between heads, which results in a Hierarchical multI-head atteNtion driven Transformer model, termed HINT, for image restoration. HINT contains two modules, i.e., the Hierarchical Multi-Head Attention (HMHA) and the Query-Key Cache Updating (QKCU) module, to address the redundancy problem that is rooted in vanilla MHA. Specifically, HMHA extracts diverse contextual features by employing heads to learn from subspaces of varying sizes and containing different information. Moreover, QKCU, comprising intra- and inter-layer schemes, further reduces the redundancy problem by facilitating enhanced interactions between attention heads within and across layers. Extensive experiments are conducted on 12 benchmarks across 5 image restoration tasks, including low-light enhancement, dehazing, desnowing, denoising, and deraining, to demonstrate the superiority of HINT. The source code is available in the supplementary materials.

Distantly Supervised Relation Extraction (DSRE) focuses on mining relational information between entities from text, which is a critical step in constructing and expanding knowledge graphs. However, this task typically faces the problem of long-tail distribution of relationships. Furthermore, much previous research has been limited to specific relational features, overlooking the potential discriminative information in the dependencies among these features. To address these issues, we propose a Type Affinity Network (TAN) that explicitly models the dependencies among relational features. Using differentiated entity-type information and local context information, we extract high-quality feature representations. The dynamic integration of type dependency information via a type affinity matrix effectively leverages potential discriminative information, enhancing relation extraction accuracy. Furthermore, aggregating implicit features of relationships forms base points as references, improving the ability of the model to handle long-tail relations. These relationship base points differ from traditional hierarchical structures, providing a more general method. We conduct extensive experiments on two benchmark DSRE datasets, namely NYT-10 and WIKI-20 m. Compared with multiple state-of-the-art DSRE methods, the TAN not only shows significant performance improvements in handling long-tail relations, but also demonstrates overall performance improvement.

In recent years, large language models (LLMs) have achieved remarkable progress in dialogue generation. However, existing automatic evaluation methods still face challenges in diverse scenarios, particularly in terms of limited generalization and low alignment with human judgment. To address these issues, we propose a paraphrase-based evaluation framework that integrates Abstract Meaning Representation (AMR) with general-purpose language models (GPT). This approach generates diverse paraphrases across lexical, syntactic and stylistic dimensions to enhance the coverage of traditional evaluation metrics. Experimental results show that incorporating paraphrase augmentation significantly improves the correlation between automatic metrics and human evaluation on multiple datasets. Additionally, extensive experiments on six mainstream LLMs demonstrate the effectiveness and generalizability of the proposed method. This study offers new insights into improving the human alignment of automatic evaluation and lays a foundation for the application and optimization of LLMs in open-domain dialogue systems.

For near-field Synthetic Aperture Interferometric Radiometer (SAIR) systems, imaging distance plays an important role in the imaging reconstruction. However, most current SAIR imaging algorithms treat the imaging distance as an exact known value, which may have a serious influence on imaging reconstruction quality caused by the misestimate of the practical imaging distance. In this letter, we propose a distance adaptive near-field SAIR imaging algorithm based on deep learning approach to handle this problem. The Complex Correlation Operation Block (CCOB) is proposed to predict the distance parameter and reconstruct the image firstly. And the residual attention enhanced module is utilized to further improve the imaging quality. The simulation results indicate that the proposed algorithm can accurately predict the imaging distance and achieve high-precision image reconstruction at different distances.