International Journal of Intelligent Systems最新文献

Brain tumors cause death to a lot of people globally. Brain tumor disease is seen as one of the most lethal diseases since its mortality rate is high. Nevertheless, this rate can be diminished if the disease is identified and treated early. Recently, healthcare providers have relied on computed tomography (CT) scans and magnetic resonance imaging (MRI) in their diagnosis. Currently, various artificial intelligence (AI)-based solutions have been implemented to diagnose this disease early to prepare suitable treatment plans. In this article, we propose a novel self-attention transfer adaptive learning approach (SATALA) to identify brain tumors. This approach is an automated AI-based model that contains two deep-learning technologies to determine the existence of brain tumors. In addition, the proposed approach categorizes the identified tumors into two groups, which are benign and malignant. The developed method incorporates two deep-learning technologies: a convolutional neural network (CNN), which is VGG-19, and a new UNET network architecture. This approach is trained and evaluated on six public datasets and attained exquisite results. It achieved an average of 95% accuracy and an F1-score of 96.61%. The proposed approach was compared with other state-of-the-art models that were reported in the related work. The conducted experiments show that the proposed approach generates exquisite outputs and exceeds other works in some scenarios. In conclusion, we can infer that the proposed approach provides trustworthy identifications of brain cancer and can be applied in healthcare facilities.

Gravitational Search Algorithm (GSA) is a well-known physics-based meta-heuristic algorithm inspired by Newton’s law of universal gravitation and performs well in solving optimization problems. However, when solving high-dimensional optimization problems, the performance of GSA may deteriorate dramatically due to severe interference of redundant dimensional information in the high-dimensional space. To solve this problem, this paper proposes a Manifold-Guided Gravitation Search Algorithm, called MGGSA. First, based on the Isomap, an effective dimension extraction method is designed. In this mechanism, the effective dimension is extracted by comparing the dimension differences of the particles located in the same sorting position both in the original space and the corresponding low-dimensional manifold space. Then, the gravitational adjustment coefficient is designed, so that the particles can be guided to move in a more appropriate direction by increasing the effect of effective dimension, reducing the interference of redundant dimension on particle motion. The performance of the proposed algorithm is tested on 35 high-dimensional (dimension is 1000) benchmark functions from CEC2010 and CEC2013, and compared with eleven state-of-art meta-heuristic algorithms, the original GSA and four latest GSA’s variants, as well as three well-known large-scale global optimization algorithms. The experimental results demonstrate that MGGSA not only has a fast convergence rate but also has high solution accuracy. Besides, MGGSA is applied to three real-world application problems, which verifies the effectiveness of MGGSA on practical applications.

The simultaneous use of multiple drugs, known as polypharmacy, heightens the risks of harmful side effects due to drug-drug interactions. Predicting these interactions is crucial in drug research due to the rising prevalence of polypharmacy. Researchers employ a graphical structure to model these interactions, representing drugs and side effects as nodes and their interactions as edges. This creates a multipartite graph that encompasses various interactions such as protein-protein interactions, drug-target interactions, and side effects of polypharmacy. In this study, a method named PU-GNN, based on graph neural networks, is introduced to predict drug side effects. The proposed method involves three main steps: (1) drug features extraction using a novel biclustering algorithm, (2) reducing uncertainity in input data using a positive-unlabeled learning algorithm, and (3) prediction of drug’s polypharmacies by utilizing a graph neural network. Performance evaluation using 5-fold cross-validation reveals that PU-GNN surpasses other methods, achieving high scores of 0.977, 0.96, and 0.949 in the AUPR, AUC, and F1 measures, respectively.

We propose a novel model-free unsupervised learning paradigm to tackle the unfavorable prevailing problem of real-world image deraining, dubbed MUL-Derain. Beyond existing unsupervised deraining efforts, MUL-Derain leverages a model-free Multiscale Attentive Filtering (MSAF) to handle multiscale rain streaks. Therefore, formulation of any rain imaging is not necessary, and it requires neither iterative optimization nor progressive refinement operations. Meanwhile, MUL-Derain can efficiently compute spatial coherence and global interactions by modeling long-range dependencies, allowing MSAF to learn useful knowledge from a larger or even global rain region. Furthermore, we formulate a novel multiloss function to constrain MUL-Derain to preserve both color and structure information from the rainy images. Extensive experiments on both synthetic and real-world datasets demonstrate that our MUL-Derain obtains state-of-the-art performance over un/semisupervised methods and exhibits competitive advantages over the fully-supervised ones.

While high-speed maglev trains offer convenient travel options, they also pose challenging issues for fault detection and early warning in critical components. This study proposes a Temporal-Knowledge fusion Spatiotemporal Graph Convolutional Network (TK-STGCN) for early warning of faults in the traction control system (TCS). Compared with the existing literature that leverages the spatiotemporal characteristics of big data for fault feature discovery, TK-STGCN focuses on integrating prior knowledge to capture correlations between data and fault mechanisms, thereby improving data processing efficiency. This requires our method not only to extract spatiotemporal features from time series but also to efficiently integrate knowledge representations with time series as inputs to the model. Specifically, structural analysis (SA) is first employed to construct the predefined structural graph for the TK-STGCN backbone network. Subsequently, a knowledge fusion unit is used to integrate the knowledge graph representation with monitoring time series data as input for the TK-STGCN model. Finally, the TK-STGCN method is applied to provide early warnings for six common faults in TCS. Analysis based on 21,498 hardware-in-the-loop experiments reveals that this method can achieve a fault warning rate of over 90%. This demonstrates that the proposed method can effectively predict faults before they occur, preventing excessive equipment damage and even catastrophic consequences.

The critical aspects of risk management include hazard identification, risk assessment, and risk control. Timely risk management is critical to company decision-making, but the process of acquiring risk management knowledge is often time-consuming and labor-intensive. Knowledge graph question answering (KGQA) provides an effective solution by delivering knowledge through accurate reasoning. However, existing KGQA methods do not cover the critical risk management aspects and are difficult to retrieve quickly and accurately from large knowledge graphs. This study describes a complex question answering method for intelligently generating risk management knowledge, specifically through multi-intent information retrieval based on knowledge subgraphs. The proposed method comprises three main modules. First, in the question understanding module, we propose an intent recognition method that integrates topic entity extraction with convolutional neural networks (CNNs) to identify eleven different user intents. To enhance the retrieval efficiency, we propose a hierarchical knowledge-embedding subgraph constructed based on company and hazard descriptions. Once user intent is identified, the information retrieval module based on a novel approximate nearest neighbor (ANN) algorithm achieves deep semantic feature matching of company and hazard expressions from the knowledge embedding subgraph. After obtaining these two deep semantic features, in the answer generation module, we propose a rule-based knowledge subgraph reasoning method to answer complex questions including single-hop, multihop, constraints, and numerical calculations. On the real risk management dataset, the precision of the intent recognition module reaches 91.3% and the information retrieval module spends only 0.36 ms, verifying that the model outperforms the existing state-of-the-art models. Meanwhile, a question answering system based on the proposed method is developed to acquire risk management knowledge: Xiao An. Compared to the popular search engine and expert system for acquiring knowledge, Xiao An achieves the best results regarding ease of use, time spent, and overall performance.

The objective of the present investigation is to introduce a novel adaptive fractional-order proportional-integral-derivative controller, which is characterized by the online tuning of its parameters by utilizing five distinct multilayer perceptron neural networks employing the extended Kalman filter. Utilizing the backpropagation algorithm in training a multilayer perceptron neural network is deemed effective in identifying the structural system and estimating the plant. The controller is applied using the Jacobian derived from the online estimated model. The utilization of adaptive interval type-2 fuzzy neural networks in conjunction with the extended Kalman filter tuning method and feedback error learning strategy results in enhanced stability and robustness of the controller in the face of estimation error, seismic disturbances, and unknown nonlinear functions. The study aims to validate the efficacy of the proposed controller by examining its performance on a 20-story nonlinear building. The numerical results show that including a compensator enhances the performance of the adaptive fractional-order proportional-integral-derivative controller. The results show that the proposed adaptive fractional-order proportional-integral-derivative controller has a better performance than other controllers and that the interstory drift ratio criterion under the El Centro earthquake with a magnitude of 1.5 times experienced an improvement of up to 65% compared to other controllers, and this amount in the Kobe earthquake reached more than 58%. Other criteria have also experienced significant improvement using the proposed controller.

In many practical communication environments, traditional feature extraction methods in spectrum sensing fail to fully exploit the information of primary users. Additionally, conventional machine learning methods have weak learning capabilities, making it difficult to maintain efficient and stable spectrum sensing performance in complex noise environments. Furthermore, non-Gaussian noise can significantly affect the detection performance of spectrum sensing. To address these issues, this paper first proposes a feature extraction method based on Hierarchical Fuzzy Dispersion Entropy (HFDE) to better extract high-frequency and low-frequency information from signal samples, providing more comprehensive features for subsequent models to optimize feature extraction effectiveness. Then, a parallel model combining Convolutional Neural Networks (CNN) with Gated Recurrent Units (GRU) is constructed to enhance learning ability. While CNN extracts local features, GRU processes temporal relationships, and the features output by both are concatenated to achieve effective feature learning and temporal modeling of primary user signal data represented by HFDE. Finally, using the feature vectors output by the CNN-GRU model, detection statistics and detection thresholds for spectrum sensing are constructed for online detection. Simulation results validate the effectiveness and robustness of this method in spectrum sensing under non-Gaussian noise. In the presence of significant non-Gaussian noise intensity and a signal-to-noise ratio of −14 dB, the detection probability can reach 97.1%. Additionally, for the detection of unknown signals, the model can still maintain a detection probability of over 90%.

Exploring and modeling the spreading process of rumors have shown great potential in improving rumor detection performance. However, existing propagation-based rumor detection models often overlook the uncertainty of the underlying propagation structure and typically require a large amount of labeled data for training. To address these challenges, we propose a novel rumor detection framework, namely, the Uncertainty-Inference Contrastive Learning (UICL) model. Specifically, UICL innovatively incorporates an edge-wise augmentation strategy into the general contrastive learning framework, including an edge-inference augmentation component and an EdgeDrop augmentation component, which primarily aim to capture the edge uncertainty of the propagation structure and alleviate the sparsity problem of the original dataset. A new negative sampling strategy is also introduced to enhance contrastive learning on rumor propagation graphs. Furthermore, we use labeled data to fine-tune the detection module. Our experiments, conducted on three real-world datasets, demonstrate that UICL can not only significantly improve detection accuracy but also reduce the dependency on labeled data compared to state-of-the-art baselines.

Addressing challenges in infrared imaging, such as low contrast, blurriness, and detail scarcity due to environmental limitations and the target’s limited radiative capacity, this research introduces a novel infrared image enhancement approach using low-light image guidance. Initially, the Cbc-SwinIR model (coordinate-based convolution- image restoration using Swin Transformer) is applied for super-resolution reconstruction of both shimmer and infrared images, improving their resolution and clarity. Next, the MAXIM model (multiaxis MLP for image processing) enhances the visibility of low-light images under low illumination. Finally, the AILI (adaptive infrared and low-light)-fusion algorithm fuses the processed low-light image with the infrared image, achieving comprehensive visual enhancement. The enhanced infrared image exhibits significant improvements: a 0.08 increase in fractal dimension (FD), 0.094 rise in information entropy, 0.00512 elevation in mean square error (MSE), and a 12.206 reduction in peak signal-to-noise ratio (PSNR). These advancements in FD and information entropy highlight a substantial improvement in the complexity and diversity of the infrared image’s features. Despite a decrease in PSNR and an increase in MSE, this indicates that the newly introduced information enhances contrast and enriches texture details in the infrared images, resulting in pixel-level variations. This methodology demonstrates considerable improvements in visual content and analytical value, proving relevant, innovative, and efficient in infrared image enhancement with broad application prospects.