Recovering potential user preferences from user-item interaction matrices is a key challenge in recommender systems. While diffusion models can sample and reconstruct preferences from latent distributions, they often fail to capture similar users' collective preferences effectively. Additionally, latent variables degrade into pure Gaussian noise during the forward process, lowering the signal-to-noise ratio, which in turn degrades performance. To address this, we propose S-Diff, inspired by graph-based collaborative filtering, better to utilize low-frequency components in the graph spectral domain. S-Diff maps user interaction vectors into the spectral domain and parameterizes diffusion noise to align with graph frequency. As a result, this anisotropic diffusion retains significant low-frequency components, preserving a high signal-to-noise ratio. S-Diff further employs a conditional denoising network to encode user interactions, recovering true preferences from noisy data. This method achieves promising results across multiple datasets.

Detecting completeness in natural language (NL) requirements traditionally involves domain modeling to establish reference models, which requires substantial manual effort and proves to be time- and labor-intensive. Large Language Models (LLMs), with an extensive knowledge base and advanced capabilities for contextual understanding and reasoning, present an opportunity for automated completeness detection in NL requirements. To better complete this task, we first propose the CMD model for implied completeness, which formally characterizes the essential nature of external completeness in NL requirements. We further establish formal definitions for requirements compositions and incompleteness dimensions, providing a qualitative analytical foundation for completeness detection. We then present ReCompGPT (Requirements Completing GPT), an LLM-powered method for detecting and repairing requirements completeness. Grounded in the proposed definitions, we construct a fine-grained chain-of-thought augmented by an action diffusion approach that provides the model with multi-perspective insights into requirements. It eliminates the need for a reference model, verifying the completeness of requirements specifications from natural language to natural language (NL2NL). Simultaneously, we develop a new dataset dedicated to detecting the external completeness in NL requirements, on which our method achieves an overall detection success rate of 93.8%, significantly enhancing performance relative to the vanilla prompts and the vanilla chain-of-thought, respectively. Overall, ReCompGPT demonstrates superior capabilities in both simple scenarios and complex cases characterized by implied requirements incompleteness. Our method proves to be effective and flexible.

The propagation of information between nodes forms the foundation for social network collaboration. Identifying influential nodes to facilitate propagation in networks is essential for downstream tasks, such as social recommendations, scientific collaborations, and epidemic control. While previous research has explored most indicators associated with the topological structure of nodes to address this issue, little attention has been given to the fact that nodes have different ranges of influence. Thus, using predetermined ranges as the basis for measuring influence may not be the most suitable approach. Therefore, this paper introduces a method that utilizes graph neural networks to aggregate information from different hop neighborhoods. It incorporates learnable parameters to measure the importance of each neighborhood. Additionally, it employs von Neumann entropy to evaluate the purity of information within the neighborhood as a regularization term, thereby enhancing the performance of the semi-supervised regression model. The effectiveness and applicability of this method have been demonstrated in various networks, exhibiting remarkable accuracy. Furthermore, it enhances the interpretability of node influence prediction.

Street view images constitute an important part of urban computing, providing data support for tasks such as autonomous driving and landscape planning, and promoting the interaction and collaboration between machines and the urban environment. However, in current practice, the usability of street view images is hindered by low-light conditions, and existing low-light enhancement methods often overlook the high-frequency characteristics specific to street views. Therefore, this paper proposes a conditional diffusion model called SVBoost that incorporates high-frequency information and color balance to achieve targeted enhancement of street view images. The proposed model demonstrates favorable performance in terms of image quality, and the enhancement effect observed in semantic segmentation tasks suggests the potential of this method for downstream applications.

With the widespread of cyber attacks, network intrusion detection system (NIDS) is becoming an important and essential tool to protect Internet of Things (IoT) environments. However, it is well known that the NIDS performance depends heavily on the effectiveness of the detection model, which can be influenced significantly by the learning mechanism and the available training data. Many existing studies try to mitigate the above challenges, but few of them consider the adaptability and the cost of deploying an NIDS, the integrity of the learning process, the capacity of model based on concrete traffic samples at the same time. To fill this gap and improve the detection performance, we propose a collaborative learning-based detection framework called ADCL, which can mitigate the limitations on the knowledge of a single model by leveraging multiple models trained in similar environments and detecting intrusions in a collaborative manner. Our evaluation results indicate that ADCL can provide better performance compared with a single model on detecting various attacks in IoT networks. Specifically, ADCL improves F-score by up to 80% for adaptability, 42% in mitigating the reliance on learning integrity, 85% for model capacity. Furthermore, the detection results of ADCL guide those single models to update and increase the F-score by 15%.

Named entity recognition (NER) is an essential subtask in natural language processing field. Recent studies have demonstrated that character-word lattice models are efficient for Chinese NER, which can leverage useful word boundary information to enhance the representation of characters. However, previous models only consider the integration of local matched word features and neglect the semantic interactions with long-range matched words. Moreover, prior methods solely achieve superficial fusion in the character-word feature space with simple methods, such as feature concatenation, but fail to implement fine-grained semantic fusion. In this paper, we propose a mixed-lattice self-attention network (MLSAN) to integrate richer word boundary information, which can explicitly capture the fine-grained correlations across characters and long-range matched words and achieve the integration of global word features. In addition, we design an end-to-end method for incorporating lexical information at the bottom layer of BERT. Compared with existing methods, our model achieves a deeper lexical knowledge fusion, which makes MLSAN perform well on cases of a small number of samples. Experimental results on four Chinese NER datasets show that our model obtains competitive performance.

With the development of airborne equipment and integrated avionics technology, the unmanned aerial vehicle (UAV) network replaces human beings in many works. However, the limited energy capacity of UAVs has great restrictions on long-time missions. To prolong the lifetime of the UAV network, we introduce the wireless static chargers (WSCs) into the UAV network, and a nondisruptive wireless rechargeable UAV network (WRUN) model is proposed, in which UAVs can be wirelessly charged without returning back to the charging platform and WSCs are scheduled to turn on and release energy only in the charging time periods. The goal of this article is to minimize the energy waste of WSCs under the premise that UAVs will not run out of energy. To calculate the efficient charging time periods of WSCs, we first discretize flight paths of UAVs so that the nondisruptive charging time schedule problem (nCTSP) that has an infinite solution space both in the spatial and time dimensions can be formalized as an optimization problem. Then, we propose the baseline algorithm exhaust candidate solutions (ECSs) to calculate the charging time periods and propose an improved algorithm using the idea of pruning (PECS) to reduce the computational complexity of ECS. Finally, experiments are conducted and PECS achieves better performance in terms of saving energy consumption and maximizing energy utilization.

Recently, wireless power transmission technology (WPT) is used in wireless rechargeable unmanned aerial vehicle (UAV) networks (WRUNs) to replenish energy for UAVs. In WRUN, if UAVs cannot replenish energy in time, the task will fail or the task completion time will be greatly prolonged. So, when and how long should the wireless static chargers (WSCs) be scheduled to replenish energy for UAVs while the energy utilization rate of WSCs can be optimized is very critical. However, the existing methods only optimize the charging time periods, that is the time durations during which WSCs release energy, ignoring the impact of the adjustable source power of WSCs. In fact, inappropriate source power will lead to the low energy utilization rate of WSCs. This paper takes both the charging time periods and the adjustable source power of WSCs into account and formulates the power-aware charging time scheduling (PACTS) problem as an optimization problem. Namely, how to decide the optimal charging time periods and the appropriate source power in each charging time period simultaneously to improve the energy utilization rate of WSCs. To solve PACTS, we first discretize the flight paths of UAVs spatiotemporally and then present a novel method to reformulate PACTS as a binary integer programming (BIP) problem that can be solved by existing algorithms. Finally, experiments are conducted to evaluate the performance of our scheme.

With the development of airborne equipment and integrated avionics technology, the unmanned aerial vehicle (UAV) network replaces human beings in many fields. To improve the durability of the UAV network, we propose a nondisruptive wireless rechargeable UAV network (WRUN) model, in which UAVs can be charged by wireless static chargers (WSCs) without returning back to the charging platform. Under the nondisruptive WRUN model, a baseline algorithm is proposed to solve the nondisruptive charging time schedule problem (nCTSP), in which chargers do not release energy all the time and can ensure UAVs do not run out of energy. Then to improve the energy utilization rate of WSCs, we propose an efficient charging time scheduling algorithm (ECTSA), in which the flight time and paths of UAVs are discretized and nCTSP is transformed into a linear binary integer programming (LBIP) problem to calculate the efficient charging time periods of WSCs. Finally, experiments are conducted to verify that ECTSA can improve the energy utilization of WSCs.

The decomposition-based algorithms [e.g., multiobjective evolutionary algorithm based on decomposition (MOEA/D)] transform a multiobjective optimization problem (MOP) into a number of single-objective optimization subproblems and solve them in a collaborative manner. It is a natural framework for using single-objective local search (LS) to solve combinatorial MOPs. However, commonly used decomposition methods, such as weighted sum (WS), Tchebycheff (TCH), and penalty-based boundary intersection (PBI) may not be good at maintaining the population diversity while providing diverse initial solutions for different LS procedures in a collaborative way. Based on our previous work on the constrained decomposition with grids (CDG), this article proposes a dynamic CDG (DCDG) framework used to design a multiobjective memetic algorithm (DCDG-MOMA). DCDG uses grids for maintaining diversity, supporting the collaborative LS. In addition, DCDG dynamically increases the number of grids for obtaining more nondominated solutions as well as the better collaborative search among them. DCDG-MOMA has been compared with several classical and state-of-the-art algorithms on multiobjective traveling salesman problem (MOTSP), multiobjective quadratic assignment problem (MOQAP), and multiobjective capacitated arc routing problem (MOCARP).