The belief-selective propagation (BsP) algorithm has recently emerged as a promising approach for massive MIMO detection. However, when applied in MIMO turbo receivers, known for their superior performance compared to separated detection and decoding (SDD) receivers, the BsP-based receiver suffers from significant performance degradation and high processing latency. To overcome these limitations, this paper proposes a fixed-constellation BsP (FCBsP) detector tailored for MIMO turbo receivers. By buildingfixed configuration setsand utilizing theapproximate multi-user interferencefor message updates, the proposed FCBsP detector achieves a better trade-off between error performance and computational complexity compared to the BsP. Furthermore, two unexplored features:information compensation and decoding-first mechanismare proposed to fine-tune the exchanged information and lower the processing latency of the FCBsP-based turbo receiver. Numerical results demonstrate that the proposed FCBsP-based turbo receiver earns about 0.7 and 1.8 dB performance gains over the BsP-based turbo receiver at BLER=10−3in an LDPC-coded 32 × 12 64-QAM MIMO system under Rayleigh and practical channels, respectively.

Low-Rank Adaptation (LoRA) has proven effective in reducing computational costs while maintaining performance comparable to fully fine-tuned foundation models across various tasks. However, its fixed low-rank structure restricts its adaptability in scenarios with substantial domain gaps, where higher ranks are often required to capture domain-specific complexities. Current adaptive LoRA methods attempt to overcome this limitation by dynamically expanding or selectively allocating ranks, but these approaches frequently depend on computationally intensive techniques such as iterative pruning, rank searches, or additional regularization. To address these challenges, we introduce Stable Rank-Guided Low-Rank Adaptation (SR-LoRA), a novel framework that utilizes the stable rank of pre-trained weight matrices as a natural prior for layer-wise rank allocation. By leveraging the stable rank, which reflects the intrinsic dimensionality of the weights, SR-LoRA enables a principled and efficient redistribution of ranks across layers, enhancing adaptability without incurring additional search costs. Empirical evaluations on few-shot tasks with significant domain gaps show that SR-LoRA consistently outperforms recent adaptive LoRA variants, achieving a superior trade-off between performance and efficiency. Our code is available at https://github.com/EndoluminalSurgicalVision-IMR/SR-LoRA.

Click-through rate (CTR) prediction is a critical task in online advertising and recommender systems, relying on effective modeling of feature interactions.Explicit interactions capture predefined relationships, such as inner products, but often suffer from data sparsity, while implicit interactions excel at learning complex patterns through non-linear transformations but lack inductive biases for efficient low-order modeling.Existing two-stream architectures integrate these paradigms but face challenges such as limited information sharing, gradient imbalance, and difficulty preserving low-order signals in sparse CTR data.We propose a novel framework, Dynamic Low-Order-Aware Fusion (DLF), which addresses these limitations through two key components: a Residual-Aware Low-Order Interaction Network (RLI) and a Network-Aware Attention Fusion Module (NAF).RLI explicitly preserves low-order signals while mitigating redundancy from residual connections, and NAF dynamically integrates explicit and implicit representations at each layer, enhancing information sharing and alleviating gradient imbalance.Together, these innovations balance low-order and high-order interactions, improving model expressiveness.Extensive experiments on public datasets demonstrate that DLF achieves

Internet of Things (IoT) devices are widely distributed with large and scattered data, risk of privacy leakage and high bandwidth cost make data transmission and aggregation impractical. Federated Learning (FL) is an important privacy-preserving multi-party learning paradigm, involving collaborative learning with others and local updating based on private data. Data heterogeneity is one of the main challenges in FL scenarios, which can greatly limit FLs applicability and performance. Most existing approaches tackle the heterogeneity challenge during local training or model aggregation while ignoring the performance drop caused by direct model aggregation and catastrophic forgetting in the global model. This work proposes a novel FL approach, called Federated Learning with Dataset Condensation (FedDC). It aims to relieve the issue of knowledge discrepancy among local models and catastrophic forgetting in the global model. Specifically, this work proposes a differential distribution matching method to summarize local data for each client without compromising on data privacy. It introduces collaborative knowledge distillation to encourage local models to learn from others. To mitigate interround forgetting, this work stabilizes the averaged global models training by leveraging auxiliary information from its immediately past one. Extensive experiments are conducted. The results show that FedDC significantly outperforms the state-ofthe-art methods on various vision tasks.

In recommendation systems, users frequently engage in multiple types of behaviors, such as clicking, adding to cart, and purchasing. Multi-behavior sequential recommendation aims to jointly consider multiple behaviors to improve the target behavior's performance. However, with diversified behavior data, user behavior sequences will become very long in the short term, which brings challenges to the efficiency of the sequence recommendation model. Meanwhile, some behavior data will also bring inevitable noise to the modeling of user interests. To address the aforementioned issues, firstly, we develop the Efficient Behavior Sequence Miner (EBM) that efficiently captures intricate patterns in user behavior while maintaining low time complexity and parameter count. Secondly, we design hard and soft denoising modules for different noise types and fully explore the relationship between behaviors and noise. Finally, we introduce a contrastive loss function along with a guided training strategy to contrast the valid information with the noisy signal in the data, and seamlessly integrate the two denoising processes to achieve a high degree of decoupling of the noisy signal. Sufficient experiments on real-world datasets demonstrate the effectiveness and efficiency of our approach in dealing with multi-behavior sequential recommendation.

Autoscaling is a critical mechanism in cloud computing, enabling the autonomous adjustment of computing resources in response to dynamic workloads. This is particularly valuable for co-located, long-running applications with diverse workload patterns. The primary objective of autoscaling is to regulate resource utilization at a desired level, effectively balancing the need for resource optimization with the fulfillment of Service Level Objectives (SLOs). Many existing proactive autoscaling frameworks may encounter prediction deviations arising from the frequent fluctuations of cloud workloads. Reactive frameworks, on the other hand, rely on realtime system feedback, but their hysteretic nature could lead to violations of stringent SLOs. Hybrid frameworks, while prevalent, often feature independently functioning proactive and reactive modules, potentially leading to incompatibility and undermining the overall decision-making efficacy. In addressing these challenges, we propose OptScaler, a collaborative autoscaling framework that integrates proactive and reactive modules through an optimization module. The proactive module delivers reliable future workload predictions to the optimization module, while the reactive module offers a self-tuning estimator for real-time updates. By embedding a Model Predictive Control (MPC) mechanism and chance constraints into the optimization module, we further enhance its robustness. Numerical results have demonstrated the superiority of our workload prediction model and the collaborative framework, leading to over a 36% reduction in SLO violations compared to prevalent reactive, proactive, or hybrid autoscalers. Notably, OptScaler has been successfully deployed at Alipay, providing autoscaling support for the world-leading payment platform.

Social networks are an essential component of the Internet of People (IoP) and play an important role in stimulating interactive communication among people. Graph convolutional networks provide methods for social network analysis with its impressive performance in semi-supervised node classification. However, the existing methods are based on the assumption of balanced data distribution and ignore the imbalanced problem of social networks. In order to extract the valuable information from imbalanced data for decision making, a novel method named minority-weighted graph neural network (mGNN) is presented in this article. It extends imbalanced classification ideas in the traditional machine learning field to graph-structured data to improve the classification performance of graph neural networks. In a node feature aggregation stage, the node membership values among nodes are calculated for minority nodes’ feature aggregation enhancement. In an oversampling stage, the cost-sensitive learning is used to improve edge prediction results of synthetic minority nodes, and further raise their importance. In addition, a Gumbel distribution is adopted as an activation function. The proposed mGNN is evaluated on six social network data sets. Experimental results show that it yields promising results for imbalanced node classification.

Abstract Question generation (QG) refers to the automatic generation of questions based on the given passages and answers, and has a wide range of application scenarios in human–computer interaction, education, medical and other fields. However, for Chinese QG, due to the lack of word separation in the writing rules of Chinese text, many methods are not suitable for it, and the generated results have incorrect word order and invalid expressions. In addition, traditional models only use labeled data, but it is difficult and expensive to obtain data labels. In order to solve this problem, this article proposes a semi‐supervised QG model termed virtual stroke‐aware copy network (VSAC Net). It is based on virtual adversarial training and can be used for Chinese QG tasks with few labeled samples. The VSAC Net model combines word embedding virtual counter disturbance and attention virtual counter disturbance, the fitting of the input layer and the attention layer is taken into account, and reduces the overfitting of the model. According to Dureader dataset, a small sample QG dataset is constructed, and the VSAC Net is used for solving. The results show that the proposed model can achieve a better generation effect on small sample datasets.

Abstract The problem of imbalanced data classification has become a research hotspot in the field of machine learning. Fuzzy support vector machine (FSVM) is an imbalanced classification processing method based on cost‐sensitive theory. The existing methods have cost‐sensitive, causing the prior distribution estimation of data inaccurate. This article proposes a novel FSVM algorithm based on the kernel local outlier factor (KLOF‐FSVM) for this problem. KLOF calculates the local outlier factor of the sample in the kernel space and assigns an appropriate membership value to the sample. This process enables the algorithm to obtain the distribution information of the data better. Compared with the algorithm based on distance only, KLOF has better robustness. It can expand the value range of majority class samples' membership degree to better balance the important relation between the minority class and the majority class. We selected some datasets in the Keel data repository and used cross‐validation to obtain the algorithm's effect under different evaluation indexes such asG‐Mean,F1 measure, and area under curve. By comparing with other algorithms, preliminary results show that this method has better classification quality.

Abstract With the continuous development of convolutional neural networks (CNNs), image classification technology has entered a new stage in solving visual cognition tasks. Recent advances have shown that if containing short connections, convolutional networks can be more accurate and efficient to train. However, these short connections almost only exist between convolutional layers, which potentially make the feature information flow insufficiently. In addition, the multi‐scale representation ability of CNNs using short connections can be further explored. Thus, the dense short connection network (DSCNet) for efficient image classification is designed in this paper. In DSCNet, we propose a simple yet effective architectural unit, namely dense short connection (DSC) module, which allows hierarchical dense short connections for multi‐scale context information across different feature channels within a single convolutional layer. DSCNets have several noteworthy advantages: they improve the multi‐scale representation ability, enhance feature propagation, and reduce the number of parameters. To validate our DSC, we conduct comprehensive experiments on CIFAR‐100 and Tiny ImageNet datasets. Experimental results show that DSCNets achieve very competitive results with previous baseline models, whilst requiring less computation to achieve higher recognition accuracy. Further ablation studies show that our proposed method can achieve consistent performance gains in image classification tasks.