Multi-head Attention Mechanism Research Articles

Prediction of PM2.5 Concentration by Transformer Model Based on Attention Mechanism

An improved transformer model based on a multi-head attention mechanism was constructed for long-term prediction of PM2.5 concentration. The monitoring data and meteorological data from 12 air monitoring stations in Beijing from March 2013 to December 2016 were collected and used in the transformer model. The Pearson coefficient was used to explore the key factors affecting PM2.5 concentration. A convolutional neural network model （ResNet50） and long short-term memory network model （LSTM） were introduced for comparison and explanatory variance （EVS）, coefficient of determination （R2）, mean square error （MSE）, and mean absolute error （MAE） were selected to evaluate the performance of the model. The Pearson coefficient results showed that PM10, SO2, and NO2, CO, and atmospheric pressure （PRES） were highly correlated with PM2.5 concentration, and dew point temperature （DEWP） was strongly correlated with PM2.5 concentration, which was consistent with the preference setting of the model's automatic screening. The MSE and R2 of the transformer model are 0.009 μg·m-3 and 0.925, respectively, which decreased by 91.09% and 30.77% of MSE and increased by 38.05% and 4.65% of R2, respectively, when compared with ResNet50 and LSTM. The transformer model could capture short-term pollution changes caused by sudden changes in meteorological conditions and long-term trends with significant seasonal changes. The fitting effect of the transformer was excellent among several models, providing a novel method for long-term prediction of PM2.5 concentration. In addition, ablation experiments revealed that the increase in R2 of the transformer was relatively small after data input or manually setting preferences, with only a 2.31% and 1.51% increase, respectively, indicating that the transformer model had strong anti-interference ability for PM10 homologous data.

Integrating Convolutional Attention and Encoder–Decoder Long Short-Term Memory for Enhanced Soil Moisture Prediction

Soil moisture is recognized as a crucial variable in land–atmosphere interactions. This study introduces the Convolutional Attention Encoder–Decoder Long Short-Term Memory (CAEDLSTM) model to address the uncertainties and limitations inherent in traditional soil moisture prediction methods, especially in capturing complex temporal dynamics across diverse environmental conditions. Unlike existing approaches, this model integrates convolutional layers, an encoder–decoder framework, and multi-head attention mechanisms for the first time in soil moisture prediction. The convolutional layers capture local spatial features, while the encoder–decoder architecture effectively manages temporal dependencies. Additionally, the multi-head attention mechanism enhances the model’s ability to simultaneously focus on multiple key influencing factors, ensuring a comprehensive understanding of complex environmental variables. This synergistic combination significantly improves predictive performance, particularly in challenging climatic conditions. The model was validated using the LandBench1.0 dataset, which includes multiple high-resolution datasets, such as ERA5-land, ERA5 atmospheric variables, and SoilGrids, covering various climatic regions, including high latitudes, temperate zones, and tropical areas. The superior performance of the CAEDLSTM model is evidenced by comparisons with advanced models such as AEDLSTM, CNNLSTM, EDLSTM, and AttLSTM. Relative to the traditional LSTM model, CAEDLSTM achieved an average increase of 5.01% in R2, a 12.89% reduction in RMSE, a 16.67% decrease in bias, and a 4.35% increase in KGE. Moreover, it effectively addresses the limitations of traditional deep learning methods in challenging climates, including tropical Africa, the Tibetan Plateau, and Southeast Asia, resulting in significant enhancements in predictive accuracy within these regions, with R2 values improving by as much as 20%. These results underscore the capabilities of CAEDLSTM in capturing complex soil moisture dynamics, demonstrating its considerable potential for applications in agriculture and water resource monitoring across diverse climates.

A residual GRU method with deep cross fusion for Alzheimer’s disease progression prediction using missing variable-length time series data

Alzheimer’s disease (AD) is a neurodegenerative disorder with a long prodromal phase. Early prediction of AD progression is crucial for improving clinical diagnosis. However, missing data and variable-length of time series data make it difficult to predict the progression of AD in clinical practice. Regarding this problem, existing researches mainly employed LSTM (Long Short-Term Memory) networks to impute missing data and make predictions, but those methods only encoded fixed-length time series data and suffered from the error accumulation, resulting in relatively unsatisfactory prediction results. To address this issue, this paper develops a novel method, named residual sharing GRU (Gated Recurrent Unit) with enhanced deep cross fusion module (RGRU-EDCF), to perform AD progression prediction. Specifically, we first design an enhanced cross deep module to impute missing data and learn complementary information. Moreover, we design a residual sharing GRU model to realize the input of variable-length time series data, thus, enhancing flexibility of model. Besides, the residual structure in the residual sharing GRU model is used for reducing error accumulation. We also add a local multi-head attention mechanism for specific time series features learning for classification and regression prediction. Finally, we use specific time series features to realize prediction of classification and regression tasks. Unlike existing methods that trained the missing data imputation and prediction model separately, our method considers an end-to-end training strategy to train both enhanced cross deep module and the residual sharing GRU model to further promote the performance of AD progression prediction. The proposed method is validated using time series data from the ADNI dataset. The accuracy can reach 95.17% in the prediction task, MAE can reach 2.64 in cognitive score prediction, and 4.04 in MRI biomarkers prediction, it is competitive and superior over other state-of-the-art methods. We also validate our proposed method on MIMIC-III dataset.

BHC-YOLOV8 : improved YOLOv8-based BHC target detection model for tea leaf disease and defect in real-world scenarios

IntroductionThe detection efficiency of tea diseases and defects ensures the quality and yield of tea. However, in actual production, on the one hand, the tea plantation has high mountains and long roads, and the safety of inspection personnel cannot be guaranteed; on the other hand, the inspection personnel have factors such as lack of experience and fatigue, resulting in incomplete and slow testing results. Introducing visual inspection technology can avoid the above problems.MethodsFirstly, a dynamic sparse attention mechanism (Bi Former) is introduced into the model backbone. It filters out irrelevant key value pairs at the coarse region level, utilizing sparsity to save computation and memory; jointly apply fine region token to token attention in the remaining candidate regions. Secondly, Haar wavelets are introduced to improve the down sampling module. By processing the input information flow horizontally, vertically, and diagonally, the original image is reconstructed. Finally, a new feature fusion network is designed using a multi-head attention mechanism to decompose the main network into several cascaded stages, each stage comprising a sub-backbone for parallel processing of different features. Simultaneously, skip connections are performed on features from the same layer, and unbounded fusion weight normalization is introduced to constrain the range of each weight value.ResultsAfter the above improvements, the confidence level of the current mainstream models increased by 7.1%, mAP0.5 increased by 8%, and reached 94.5%. After conducting ablation experiments and comparing with mainstream models, the feature fusion network proposed in this paper reduced computational complexity by 10.6 GFlops, increased confidence by 2.7%, and increased mAP0.5 by 3.2%.DiscussionThis paper developed a new network based on YOLOv8 to overcome the difficulties of tea diseases and defects such as small target, multiple occlusion and complex background.

Data-driven forecasting framework for daily reservoir inflow time series considering the flood peaks based on multi-head attention mechanism

Accurate and reliable daily reservoir inflow forecast plays an essential role in several applications involving the management and planning of water resources, such as hydroelectric generation, flood control, water supply, and basin ecological dispatching. Runoff usually exhibits strong non-linearity, high uncertainty, and spatial and temporal variability. Existing techniques fail to capture complete dynamics change processes effectively. A data-driven forecasting framework for daily reservoir inflow time series considering the flood peaks based on a multi-head attention mechanism was developed, referred to as the GWOCS-VMD-CNN-Transformer (GCVCT). First, the model utilize Grey Wolf Optimizer coupled with Cuckoo Search (GWO-CS) algorithms to optimize parameters in variational mode decomposition model (VMD). This approach helps obtain highly correlated intrinsic mode function (IMF) components, enhancing the frequency resolution of the input dataset. The proposed method overcomes the bottleneck of other available methods by decomposing the time series to capture the main long-term and short-term properties of hydrological processes. Second, the convolution neural network and Transformer (CNN-Transformer) are based on a multi-head attention mechanism as the objective predictive method. Finally, six evaluation indicators verify the performance of the proposed approach. The approach’s reliability was evaluated using the historical daily reservoir inflow data from the Xiluodu (XLD) and Wudongde (WDD) reservoirs in the Jinsha River Basin, China. Several single and hybrid models were developed for comparative analysis. The results indicate that the proposed ensemble approach fits better than other developed model methods. The GCVCT model showed excellent performance in forecasting the inflows of XLD and WDD reservoirs, with NSE values of 0.985 and 0.984, respectively. Furthermore, the GCVCT framework forecast capacity for peak inflow was further verified through discussion and analysis of the 48 peak flows during the validation period, consistently outperforming other models in predicting peak flow for both study reservoirs. This framework provides an effective method for the scientific optimal scheduling of hydropower reservoirs, enabling more sustainable and efficient management practices. It also demonstrates the potential of powerful deep-learning models in intelligent hydrological forecasting.

Bearing fault diagnosis based on sparsity structure pruning graph attention network

Abstract Graph neural networks have been widely used in the field of bearing fault diagnosis, which can deal with non-Euclidean space data and dig deep the relationship between signals. However, most graph neural networks do not distinguish the importance of nodes in information aggregation, and do not take edge noise and data redundancy into account when constructing the graph structure, which affects the diagnostic accuracy. To solve these problems, a fault diagnosis method of graph attention network based on sparsity structure pruning is proposed. Firstly, a sparsity coefficient is introduced to construct the graph structure, and pruning operations are carried out according to the coefficient and the weight of the edges to avoid invalid fusion of information. Then, a graph attention network model based on sparsity structure pruning is constructed, and features of different scales are aggregated into new node representations through multi-head attention mechanism. Finally, the fault diagnosis of bearing is carried out according to the extracted signal discrimination characteristics. To verify the effectiveness of the proposed method, experiments are performed on two different fault diagnosis datasets and compared with other graph neural network methods. The results show that the accuracy and stability of the proposed method are superior to other methods even under the condition of low signal to noise ratio (SNR).

Enhancing text classification through novel deep learning sequential attention fusion architecture

<span lang="EN-US">Text classification is a pivotal task within natural language processing (NLP), aimed at assigning semantic labels to text sequences. Traditional methods of text representation often fall short in capturing intricacies in contextual information, relying heavily on manual feature extraction. To overcome these limitations, this research work presents the sequential attention fusion architecture (SAFA) to enhance the features extraction. SAFA combines deep long sort-term memory (LSTM) and multi-head attention mechanism (MHAM). This model efficiently preserves data, even for longer phrases, while enhancing local attribute understanding. Additionally, we introduce a unique attention mechanism that optimizes data preservation, a crucial element in text classification. The paper also outlines a comprehensive framework, incorporating convolutional layers and pooling techniques, designed to improve feature representation and enhance classification accuracy. The model's effectiveness is demonstrated through 2-dimensional convolution processes and advanced pooling, significantly improving prediction accuracy. This research not only contributes to the development of more accurate text classification models but also underscores the growing importance of NLP techniques.</span>

A GCN-LSTM framework for link prediction in dynamic SIoT networks

The Social Internet of Things (SIoT) paradigm combines the Internet of Things (IoT) with social networking principles, enabling autonomous device interactions. However, the dynamic, heterogeneous, and temporal nature of SIoT networks presents significant challenges for link prediction. While existing models, such as SIoTPredict (Aljubairy et al., 2020), exploits social network analysis and machine learning, they primarily focus only on certain aspects of social relationships between devices, lacking consideration of temporal dynamics, environmental factors, and device heterogeneity. In contrast, we propose a novel framework that integrates Graph Convolutional Networks (GCNs) with Long Short-Term Memory (LSTM) networks and a Multi-Head Attention mechanism, further enhanced by a dynamic feature scheduling strategy, to accurately capture the spatial and temporal dependencies in SIoT environments. Additionally, our work introduces a MATLAB-based SIoT Simulator, providing a robust platform for generating realistic dynamic scenarios and evaluating complex Social Object Relationships. Unlike SIoTPredict, which is optimized for real-time applications but focuses narrowly on social aspects, our approach incorporates a broader range of factors, enabling better predictive accuracy in scenarios with high temporal variability. Extensive co-simulations with MATLAB and PyTorch demonstrate that our framework significantly outperforms existing methods in terms of predictive accuracy, adaptability, and robustness. By bridging critical gaps in current methodologies, this work establishes a comprehensive foundation for future research and practical applications in dynamic SIoT networks.

A non-functional requirements classification model based on cooperative attention mechanism fused with label embedding

Intelligent classification of software requirements is a hot research issue in the field of requirements engineering. Complete and accurate identification of functional requirements (FRs) and non-functional requirements (NFRs) is the primary task of requirements engineering. However, in real software projects, NFRs are easily neglected and may become a potential risk of project failure. Text is the main source of information about software requirements. With the increasing scale of software projects, a large number of complex types of text materials are used for software requirements analysis. Manual identification of NFRs of software projects has the problems of easy omission, ambiguity, vagueness, and high time-consuming cost. Based on the above existing defects, a deep neural network model named CAFLE is designed in this paper to solve it. CAFLE is composed of two parts, Text-label cooperative attention encoder (TLCAE) and Label decoder (LD). TLCAE adopts a Bi-directional long short-term memory network (Bi-LSTM) and multi-head cooperative attention mechanism to generate an encoded representation of the mutual involvement of requirement classification labels and requirement text. LD is an LSTM decoder with an attention mechanism constructed for the multi-class classification task of requirement text. LD utilizes the representation generated by TLCAE for prediction. Experimental results on the PROMISE benchmark dataset show that CAFLE outperforms existing NFRs classification methods with an F1 score of 95%.

A cellular traffic prediction method based on diffusion convolutional GRU and multi-head attention mechanism

A cellular traffic prediction method based on diffusion convolutional GRU and multi-head attention mechanism

Fault diagnosis of photovoltaic array based on gated residual network with multi-head self attention mechanism

Fault diagnosis of photovoltaic array based on gated residual network with multi-head self attention mechanism

A Cepstrum-Informed neural network for Vibration-Based structural damage assessment

Data-driven methods for vibration-based Structural Health Monitoring (SHM) have gained significant popularity for their straightforward modeling process and real-time tracking capabilities. However, developing complex models such as deep neural networks can pose challenges, including limited interpretability and substantial computational demands, due to the large number of parameters and deep layer stacking. This study introduces a novel Cepstrum-Informed Attention-Based Network (CIABN) developed to model power cepstral coefficients of structural acceleration responses, guided by cepstrum-based physical properties to facilitate efficient structural damage assessment. The CIABN integrates three key components: a unique input–output mapping based on weighted cepstral coefficients, a novel cepstral positional encoding mechanism, and a multi-head self-attention mechanism. The unique input–output mapping enables appreciable model generalization in overall structural characteristics, with the weighted cepstral coefficients serving as informative and compact data for efficient neural network modeling. The developed cepstral positional encoding scientifically guides the model to capture the coefficient indices, and the underlying trend of cepstral coefficients primarily governed by overall structural characteristics. The multi-head attention mechanism enables computationally efficient parallel analysis of interdependencies among coefficients, facilitating the development of a lightweight network. The effectiveness and superiority of the method have been validated using both simulated and experimental structural data.

Cross-modal embedding integrator for disease-gene/protein association prediction using a multi-head attention mechanism.

Knowledge graphs, powerful tools that explicitly transfer knowledge to machines, have significantly advanced new knowledge inferences. Discovering unknown relationships between diseases and genes/proteins in biomedical knowledge graphs can lead to the identification of disease development mechanisms and new treatment targets. Generating high-quality representations of biomedical entities is essential for successfully predicting disease-gene/protein associations. We developed a computational model that predicts disease-gene/protein associations using the Precision Medicine Knowledge Graph, a biomedical knowledge graph. Embeddings of biomedical entities were generated using two different methods-a large language model (LLM) and the knowledge graph embedding (KGE) algorithm. The LLM utilizes information obtained from massive amounts of text data, whereas the KGE algorithm relies on graph structures. We developed a disease-gene/protein association prediction model, "Cross-Modal Embedding Integrator (CMEI)," by integrating embeddings from different modalities using a multi-head attention mechanism. The area under the receiver operating characteristic curve of CMEI was 0.9662 (± 0.0002) in predicting disease-gene/protein associations. In conclusion, we developed a computational model that effectively predicts disease-gene/protein associations. CMEI may contribute to the identification of disease development mechanisms and new treatment targets.

A power grid topology detection method based on edge graph attention neural network

Against the backdrop of growing power demand and market liberalization, operators optimize power transmission and system load balance through network reconfiguration by switching lines. However, the complex node connections and dynamic load variations exceed the capabilities of traditional algorithms, and data noise may cause detection errors, affecting grid dispatch and stability. To address this issue, a solution is proposed based on the Edge Graph Attention Neural Network (EGAT) model, providing an in-depth analysis of the reconfigured grid structure. The model employs a multi-head attention mechanism to integrate node and edge features from different layers, enhancing feature fusion and extracting critical topological information, thereby improving the accuracy and robustness of grid topology detection after reconfiguration. The method was tested on IEEE 14-bus, IEEE 39-bus, and IEEE 118-bus systems, achieving detection accuracies of 92.30 %, 90.14 %, and 87.25 %, respectively, significantly outperforming other neural network methods. However, the complexity of the model may result in additional computational overhead.

Fault diagnosis of rolling bearings under variable operating conditions based on improved graph neural networks

Abstract To address the issues of low diagnostic accuracy, insufficient generalization, and poor robustness in traditional fault diagnosis methods across different equipment and varying operating conditions, this paper proposes an improved graph neural network-based fault diagnosis method for rolling bearings to enhance model performance under complex conditions. First, the optimized wavelet transform coefficient features are used as nodes, and by exploring the correlations between features, node adjacency relationships are constructed. The associations between fault modes and feature node graphs under different conditions are studied, and a fault feature graph sample set based on subgraph structures is built, providing data for the subsequent graph neural network learning. Then, a multi-head attention mechanism (MHGAT) and multi-scale feature adaptive perception pooling (MSF-ASAP) are integrated to construct a multi-head graph attention mechanism model based on multi-scale feature adaptive perception pooling (MSM-GAT). MHGAT enhances the model's ability to perceive global information by learning different features from multiple perspectives and dimensions, thus improving the model's generalization. MSF-ASAP adaptively selects and aggregates information at multiple scales, enabling the model to effectively extract key features under various operating conditions, resist noise interference, and adapt to local information changes, thus improving the model's robustness under varying conditions and noisy environments. Experimental results under multiple and continuously varying conditions demonstrate that the proposed method outperforms traditional methods in terms of diagnostic accuracy and robustness. Notably, it exhibits excellent generalization when identifying unknown conditions, achieving over 95% accuracy in recognizing new conditions and maintaining over 92.5% accuracy in noisy environments.

Early stroke behavior detection based on improved video masked autoencoders for potential patients

Stroke is the prevalent cerebrovascular disease characterized by significant incidence and disability rates. To enhance the early perceive and detection of potential stroke patients, the early stroke behavior detection based on improved Video Masked Autoencoders (VideoMAE) for potential patients (EPBR-PS) is proposed. The proposed method begins with novel time interval-based sampling strategy, capturing video frame sequences enriched with sparse motion features. On the basis of establishing the masking mechanism for adjacent frames and pixel blocks within these sequences, The EPBR-PS employes pipeline mask strategy to extract spatiotemporal features effectively. Then, the local convolution attention mechanism is designed to capture local dynamic feature information, and central to the EPBR-PS is the integration of local convolutional attention mechanism with VideoMAE's multi-head attention mechanism. This integration facilitates the simultaneous leveraging of global high-level semantics and local dynamic feature information. Dual attention mechanism-based method for the fusion of these global and local features is proposed. After that, the optimal parameters of EPBR-PS were determined through the experiment of learning rate and fusion weights of different features. On the NTU-ST dataset, comparative analysis with eight models demonstrated the superiority of EPBR-PS, evidenced by the average recognition accuracy of 89.61%, surpassing that 1.67% over the benchmark VideoMAE. On the HMDB51 dataset, EPBR-PS has Top1 of 71.31%, which is 0.73% higher than that of the VideoMAE, providing the viable behavior detection for perception early signs of potential stroke in the home environment. This code is available at https://github.com/wang-325/EPBR-PS/.

Dynamic inferences of crash risks in freeway merging zones: a spatio-temporal deep learning model

Freeway merging zones are critical for freeway operations and management due to potential crashes arising from complex vehicle merging behaviours. This paper investigates the application of a spatio-temporal deep learning model to infer crash risks in the zones. We first introduce a crash risk index based on Time-To-Collision and vehicle merging patterns. An innovation is that the developed spatio-temporal transformer model can analyze the evolving risk index. This model effectively captures dynamic risk features through a multi-head attention mechanism within its spatio-temporal learning components. Numerical experiments on nine inference tasks with varying spatial resolutions show improved performance of the model with lower resolution. Moreover, the ST-Transformer model is benchmarked against three advanced deep learning models, which consistently demonstrates its superiority in capturing spatio-temporal dependence in risk sequences. This investigation significantly contributes to a richer understanding of proactive traffic safety, providing valuable insights for advanced freeway management and driver assistance systems.

Document-level relation extraction via dual attention fusion and dynamic asymmetric loss

Document-level relation extraction (RE), which requires integrating and reasoning information to identify multiple possible relations among entities. However, previous research typically performed reasoning on heterogeneous graphs and set a global threshold for multiple relations classification, regardless of interaction reasoning information among multiple relations and positive–negative samples imbalance on databases. This paper proposes a novel framework for Document-level RE with two techniques, dual attention fusion and dynamic asymmetric loss. Concretely, to obtain more interdependency feature learning, we construct entity pairs and contextual matrixes using multi-head axial attention and co-attention mechanism to learn the interaction among entity pairs deeply. To alleviate the hard-thresholds influence from positive–negative imbalance samples, we dynamically adjust weights to optimize the probabilities of different labels. We evaluate our model on two benchmark document-level RE datasets, DocRED and CDR. Experimental results show that our DASL (Dual Attention fusion and dynamic aSymmetric Loss) obtains superior performance on two public datasets, we further provide extensive experiments to analyze how dual attention fusion and dynamic asymmetric loss guide the model for better extracting multi-label relations among entities.

Integrated Extraction of Entities and Relations via Attentive Graph Convolutional Networks

For information security, entity and relation extraction can be applied in sensitive information protection, data leakage detection, and other aspects. The current approaches to entity relation extraction not only ignore the relevance and dependency between name entity recognition and relation extraction but also may result in the cumulative propagation of errors. To solve this problem, it is proposed that an end-to-end joint entity and relation extraction model based on the Attention mechanism and Graph Convolutional Network (GCN) to simultaneously extract named entities and their relationships. The model includes three parts: the detection of entity span, the construction of an entity relation weighted graph, and the inference of entity relation type. Firstly, the detection of entity spans is viewed as a sequence labeling problem, and a multi-feature fusion approach for word embedding representation is designed to calculate all entity spans in a sentence to form an entity span matrix. Secondly, the entity span matrix is employed in the Multi-Head Attention mechanism for constructing the weighted adjacency matrix of the entity relation graph. Finally, for the inference of entity relation type, considering the interaction between entities and relations, the entity span matrix and relation connection matrix are simultaneously fed into the GCN for integrated extraction of entities and relations. Our model is evaluated on the public NYT dataset, attaining a precision of 66.4%, a recall of 63.1%, and an F1 score of 64.7% for joint entity and relation extraction, significantly outperforming other approaches. Experiments demonstrate that the proposed model is helpful for inferring entities and relations, considering the interaction between entities and relations through the Attention mechanism and GCN.

AGCNAF: predicting disease-gene associations using GCN and multi-head attention to fuse the similarity features

Abstract Association prediction between diseases and genes is a critical step in revealing the molecular mechanisms of diseases and developing drug treatment strategies. With the explosive growth of data in the biomedical field, how to effectively utilize these data for accurate prediction has become a hotspot and challenge in current research. To overcome the limitations of current prediction methods in dealing with complex biological network structures and feature extraction, this study proposes AGCNAF, a method that combines an unsupervised Graph Convolutional Network (GCN) and a multi-head attention mechanism. The metagraph-guided random walk strategy enables AGCNAF to capture local and high-order topological structures in the graph, while GCN is responsible for realizing deep feature extraction of these structures. By incorporating similarity features through the multi-attention mechanism, AGCNAF achieves effective integration of global and local features, which significantly improves the prediction performance. By incorporating similarity features through the multi-attention mechanism, AGCNAF achieves effective integration of global and local features, which significantly improves the prediction performance. By utilizing the machine learning binary classification model for prediction, the experimental results through five-fold cross-validation show that AGCNAF demonstrates significant advantages in prediction performance compared to existing methods, with its AUC and AUPR reaching 0.9686 and 0.9709, respectively, and the AUC up to 0.9812 under specific conditions. 
To verify the practical application value of AGCNAF, this study also conduct case studies on Alzheimer's disease, lung cancer, and breast cancer. The results further confirm the excellent performance of AGCNAF in identifying potential disease-gene associations, which opens up new possibilities for future disease-gene research.