Complex & Intelligent Systems最新文献

Polysomnography is the diagnostic gold standard for obstructive sleep apnea-hypopnea syndrome (OSAHS), requiring medical professionals to analyze apnea-hypopnea events from multidimensional data throughout the sleep cycle. This complex process is susceptible to variability based on the clinician’s experience, leading to potential inaccuracies. Existing automatic diagnosis methods often overlook multimodal physiological signals and medical prior knowledge, leading to limited diagnostic capabilities. This study presents a novel heterogeneous graph convolutional fusion network (HeteroGCFNet) leveraging multimodal physiological signals and domain knowledge for automated OSAHS diagnosis. This framework constructs two types of graph representations: physical space graphs, which map the spatial layout of sensors on the human body, and process knowledge graphs which detail the physiological relationships among breathing patterns, oxygen saturation, and vital signals. The framework leverages heterogeneous graph convolutional neural networks to extract both localized and global features from these graphs. Additionally, a multi-head fusion module combines these features into a unified representation for effective classification, enhancing focus on relevant signal characteristics and cross-modal interactions. This study evaluated the proposed framework on a large-scale OSAHS dataset, combined from publicly available sources and data provided by a collaborative university hospital. It demonstrated superior diagnostic performance compared to conventional machine learning models and existing deep learning approaches, effectively integrating domain knowledge with data-driven learning to produce explainable representations and robust generalization capabilities, which can potentially be utilized for clinical use. Code is available at https://github.com/AmbitYuki/HeteroGCFNet.

Recommender systems are of increasing importance owing to the growth of social networks and the complexity of user behavior, and cater to the personalized needs of users. To improve recommendation performance, several methods have emerged and made a combination of knowledge graphs and recommender systems. However, the majority of approaches faces issues like overlooking spatiotemporal features and lacking dynamic modeling. The former restricts the flexibility of recommendations, while the latter renders recommendations unable to adapt to the changing interests of users. To overcome these limitations, a novel dynamic preference recommendation model based on spatiotemporal knowledge graphs (DRSKG), which captures preferences dynamically, is proposed in this paper. Constructed by knowledge graphs, the model integrates spatiotemporal features and takes into account the dynamic preferences of users across various temporal, spatial, and situational contexts. Therefore, DRSKG not only describes the spatiotemporal characteristics of user behaviors more accurately but also models the evolution of dynamic preferences in spatiotemporal changes. Massive experiments demonstrate that the proposed model exhibits significant recommendation enhancement compared with the traditional one, achieving up to 7% and 5% improvements in terms of Precision and Recall metrics, respectively.

Adaptive traffic signal control is the core of the intelligent transportation system (ITS), which can effectively reduce the pressure on traffic congestion and improve travel efficiency. Methods based on deep Q-leaning network (DQN) have become the mainstream to solve single-intersection traffic signal control. However, most of them neglect the important difference of samples and the dependence of traffic states, and cannot quickly respond to randomly changing traffic flows. In this paper, we propose a new single-intersection traffic signal control method (Pri-DDQN) based on reinforcement learning and model the traffic environment as a reinforcement learning environment, and the agent chooses the best action to schedule the traffic flow at the intersection based on the real-time traffic states. With the goal of minimizing the waiting time and queue length at intersections, we use double DQN to train the agent, incorporate traffic state and reward into the loss function, and update the target network parameters asynchronously, to improve the agent’s learning ability. We try to use the power function to dynamically change the exploration rate to accelerate convergence. In addition, we introduce a priority-based dynamic experience replay mechanism to increase the sampling rate of important samples. The results show that Pri-DDQN achieves better performance, compared to the best baseline, it reduces the average queue length is reduced by 13.41%, and the average waiting time by 32.33% at the intersection.

Capsule networks overcome the two drawbacks of convolutional neural networks: weak rotated object recognition and poor spatial discrimination. However, they still have encountered problems with complex images, including high computational cost and limited accuracy. To address these challenges, this work has developed effective solutions. Specifically, a novel windowed dynamic up-and-down attention routing process is first introduced, which can effectively reduce the computational complexity from quadratic to linear order. A novel deconvolution-based decoder is also used to further reduce the computational complexity. Then, a novel LayerNorm strategy is used to pre-process neuron values in the squash function. This prevents saturation and mitigates the gradient vanishing problem. In addition, a novel gradient-friendly network structure is developed to facilitate the extraction of complex features with deeper networks. Experiments show that our methods are effective and competitive, outperforming existing techniques.

Zero-shot relation extraction (ZSRE) is essential for improving the understanding of natural language relations and enhancing the accuracy and efficiency of natural language processing methods in practical applications. However, the existing ZSRE models ignore the importance of semantic information fusion and possess limitations when used for zero-shot relation extraction tasks. Thus, this paper proposes a dual contrastive learning framework and a cross-attention network module for ZSRE. First, our model designs a dual contrastive learning framework to compare the input sentences and relation descriptions from different perspectives; this process aims to achieve better separation between different relation categories in the representation space. Moreover, the cross-attention network of our model is introduced from the computer vision field to enhance the attention paid by the input instance to the relevant information of the relation description. The experimental results obtained on the Wiki-ZSL and FewRel datasets fully demonstrate the effectiveness of our approach.

Applying an error-backpropagation algorithm to spiking neural networks frequently needs to employ fictive derivatives of spike functions (popularly referred to as surrogate gradients) because the spike function is considered non-differentiable. The non-differentiability comes into play given that the spike function is viewed as a numeric function, most popularly, the Heaviside step function of membrane potential. To get back to basics, the spike function is not a numeric but a Boolean function that outputs True or False upon the comparison of the current potential and threshold. In this regard, we propose a method to evaluate the gradient of spike function viewed as a Boolean function for fixed- and floating-point data formats. For both formats, the gradient is considerably similar to a delta function that peaks at the threshold for spiking, which justifies the approximation of the spike function to the Heaviside step function. Unfortunately, the error-backpropagation algorithm with this gradient function fails to outperform popularly employed surrogate gradients, which may arise from the narrow peak of the gradient function and consequent potential undershoot and overshoot around the spiking threshold with coarse timesteps. We provide theoretical grounds of this hypothesis.

Gendered disinformation undermines women’s rights, democratic principles, and national security by worsening societal divisions through authoritarian regimes’ intentional weaponization of social media. Online misogyny represents a harmful societal issue, threatening to transform digital platforms into environments that are hostile and inhospitable to women. Despite the severity of this issue, efforts to persuade digital platforms to strengthen their protections against gendered disinformation are frequently ignored, highlighting the difficult task of countering online misogyny in the face of commercial interests. This growing concern underscores the need for effective measures to create safer online spaces, where respect and equality prevail, ensuring that women can participate fully and freely without the fear of harassment or discrimination. This study addresses the challenge of detecting misogynous content in bilingual (English and Italian) online communications. Utilizing FastText word embeddings and explainable artificial intelligence techniques, we introduce a model that enhances both the interpretability and accuracy in detecting misogynistic language. To conduct an in-depth analysis, we implemented a range of experiments encompassing classic machine learning methodologies and conventional deep learning approaches to the recent transformer-based models incorporating both language-specific and multilingual capabilities. This paper enhances the methodologies for detecting misogyny by incorporating incremental learning for cutting-edge datasets containing tweets and posts from different sources like Facebook, Twitter, and Reddit, with our proposed approach outperforming these datasets in metrics such as accuracy, F1-score, precision, and recall. This process involved refining hyperparameters, employing optimization techniques, and utilizing generative configurations. By implementing Local Interpretable Model-agnostic Explanations (LIME), we further elucidate the rationale behind the model’s predictions, enhancing understanding of its decision-making process.

Survival risk prediction models have become important tools for clinicians to improve cancer treatment decisions. In the medical field, using gene expression data to build deep survival neural network models significantly improves accurate survival prognosis. However, it still poses a challenge in building an efficient method to improve the accuracy of cancer-specific survival risk prediction, such as data noise problem. In order to solve the above problem, we propose a diversity reweighted deep survival neural network method with grid optimization (DRGONet) to improve the accuracy of cancer-specific survival risk prediction. Specifically, reweighting can be employed to adjust the weights assigned to each data point in the dataset based on their importance or relevance, thereby mitigating the impact of noisy or irrelevant data and improving model performance. Incorporating diversity into the goal of multiple learning models can help minimize bias and improve learning outcomes. Furthermore, hyperparameters can be optimized with grid optimization. Experimental results have demonstrated that our proposed approach has significant advantages (improved about 5%) in real-world medical scenarios, outperforming state-of-the-art comparison methods by a large margin. Our study highlights the significance of using DRGONet to overcome the limitations of building accurate survival prediction models. By implementing our technique in cancer research, we hope to reduce the suffering experienced by cancer patients and improve the effectiveness of treatment.

Influence Maximization (IM) stands as a central challenge within the domain of complex network analysis, with the primary objective of identifying an optimal seed set of a predetermined size that maximizes the reach of influence propagation. Over time, numerous methodologies have been proposed to address the IM problem. However, one certain network referred to as Imbalanced Heterogeneous Networks (IHN), which widely used in social situation, urban and rural areas, and merchandising, presents challenges in achieving high-quality solutions. In this work, we introduce the Lightweight Reinforcement Learning algorithm with Prior knowledge (LRLP), which leverages the Struc2Vec graph embedding technique that captures the structural similarity of nodes to generate vector representations for nodes within the network. In details, LRLP incorporates prior knowledge based on a group of centralities, into the initial experience pool, which accelerates the reinforcement learning training for better solutions. Additionally, the node embedding vectors are input into a Deep Q Network (DQN) to commence the lightweight training process. Experimental evaluations conducted on synthetic and real networks showcase the effectiveness of the LRLP algorithm. Notably, the improvement seems to be more pronounced when the the scale of the network is larger. We also analyze the effect of different graph embedding algorithms and prior knowledge on algorithmic results. Moreover, we conduct an analysis about some parameters, such as number of seed set selections T, embedding dimension d and network update frequency C. It is significant that the reduction of number of seed set selections T not only keeps the quality of solutions, but lowers the algorithm’s computational cost.

Object detection using UAV images is a current research focus in the field of computer vision, with frequent advancements in recent years. However, many methods are ineffective for challenging UAV images that feature uneven object scales, sparse spatial distribution, and dense occlusions. We propose a new algorithm for detecting small objects in UAV images, called ATBHC-YOLO. Firstly, the MS-CET module has been introduced to enhance the model’s focus on global sparse features in the spatial distribution of small objects. Secondly, the BHC-FB module is proposed to address the large-scale variance of small objects and enhance the perception of local features. Finally, a more appropriate loss function, WIoU, is used to penalise the quality variance of small object samples and further enhance the model’s detection accuracy. Comparison experiments on the DIOR and VEDAI datasets validate the effectiveness and robustness of the improved method. By conducting experiments on the publicly available UAV benchmark dataset Visdrone, ATBHC-YOLO outperforms the state-of-the-art method(YOLOv7) by 3.5%.