Applied Intelligence最新文献_第7页

Reducing oversmoothing through informed weight initialization in graph neural networks

IF 3.4 2区计算机科学

Applied Intelligence Pub Date : 2025-04-07 DOI: 10.1007/s10489-025-06426-0

Dimitrios Kelesis, Dimitris Fotakis, Georgios Paliouras

{"title":"Reducing oversmoothing through informed weight initialization in graph neural networks","authors":"Dimitrios Kelesis, Dimitris Fotakis, Georgios Paliouras","doi":"10.1007/s10489-025-06426-0","DOIUrl":"10.1007/s10489-025-06426-0","url":null,"abstract":"<div><p>In this work, we generalize the ideas of Kaiming initialization to Graph Neural Networks (GNNs) and propose a new scheme (G-Init) that reduces oversmoothing, leading to very good results in node and graph classification tasks. GNNs are commonly initialized using methods designed for other types of Neural Networks, overlooking the underlying graph topology. We analyze theoretically the variance of signals flowing forward and gradients flowing backward in the class of convolutional GNNs. We then simplify our analysis to the case of the GCN and propose a new initialization method. Results indicate that the new method (G-Init) reduces oversmoothing in deep GNNs, facilitating their effective use. Our approach achieves an accuracy of 61.60% on the <i>CS</i> dataset (32-layer GCN) and 69.24% on <i>Cora</i> (64-layer GCN), surpassing state-of-the-art initialization methods by 25.6 and 8.6 percentage points, respectively. Extensive experiments confirm the robustness of our method across multiple benchmark datasets, highlighting its effectiveness in diverse settings. Furthermore, our experimental results support the theoretical findings, demonstrating the advantages of deep networks in scenarios with no feature information for unlabeled nodes (i.e., “cold start” scenario).</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 7","pages":""},"PeriodicalIF":3.4,"publicationDate":"2025-04-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10489-025-06426-0.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143786588","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Pgcnn: an interpretable graph convolutional neural network for predicting the mechanical properties of Ti-6Al-4V alloy

IF 3.4 2区计算机科学

Applied Intelligence Pub Date : 2025-04-07 DOI: 10.1007/s10489-025-06401-9

Zihao Gao, Changsheng Zhu, Yafeng Shu, Canglong Wang, Yupeng Chen, Shaohui Wang

{"title":"Pgcnn: an interpretable graph convolutional neural network for predicting the mechanical properties of Ti-6Al-4V alloy","authors":"Zihao Gao, Changsheng Zhu, Yafeng Shu, Canglong Wang, Yupeng Chen, Shaohui Wang","doi":"10.1007/s10489-025-06401-9","DOIUrl":"10.1007/s10489-025-06401-9","url":null,"abstract":"<div><p>This study introduces a polycrystalline graph convolutional network (PGCNN) to predict the mechanical properties of Ti-6Al-4V alloy’s dual-phase polycrystalline microstructure. The model captures complex inter-grain interactions. It integrates node features and graph structural information to map microstructures to macroscopic mechanical properties. The PGCNN model demonstrated exceptional predictive performance (mean absolute relative error, MARE = 0.369%). It remained robust in handling nonlinear relationships and capturing high-order inter-grain interactions, even with limited datasets (MARE = 1.985%). We evaluated the interpretability of the PGCNN model through analyses at the node, edge, and graph structure levels, offering comprehensive insights. At the node level, the influence of each grain (node) on the output was quantified, clarifying the direct link between individual grains and macroscopic performance. Edge level analysis emphasized the importance of inter-grain interactions. It laid the groundwork for identifying grain boundaries that significantly affect mechanical properties. Graph level analysis quantified the overall impact of microstructural features on macroscopic performance. This provided insights into the complex “microstructure–mechanical property” relationship in dual-phase polycrystals.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 7","pages":""},"PeriodicalIF":3.4,"publicationDate":"2025-04-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143786586","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A digital twin-based framework for load identification using odd harmonic current plots 利用奇次谐波电流图识别负载的数字孪生框架

IF 3.4 2区计算机科学

Applied Intelligence Pub Date : 2025-04-07 DOI: 10.1007/s10489-025-06512-3

Dimitra N. Mylona, Aggelos S. Bouhouras

{"title":"A digital twin-based framework for load identification using odd harmonic current plots","authors":"Dimitra N. Mylona, Aggelos S. Bouhouras","doi":"10.1007/s10489-025-06512-3","DOIUrl":"10.1007/s10489-025-06512-3","url":null,"abstract":"<div><p>Non-intrusive Load Monitoring (NILM) techniques are becoming more and more widespread, because of the interest that consumers have in efficient energy consumption and management. At the same time, NILM application along with Demand Side Management (DSM) schemes could face Distribution Network (DN) operational issues like congestion management. The advent of Digital Twin (DT) technology offers a sustainable solution for more effective energy management in real-time applications. In addition, recent developments in NILM suggest that high sampling rates of the aggregated extracted signal could enable better performance for load disaggregation. This work explores DT integration with NILM for a real-time appliance classification scheme. More specifically, a Convolutional Neural Network (CNN) model fed with images that depict odd current harmonics is utilized to classify the appliance(s) operation. The images are extracted exploiting the high sampling measurements provided by the PLAID dataset. Three different scenarios that include various residential appliances are examined comprising both single and combined appliance operation, as well as event detection (appliance’s activation/de-activation). The results of the proposed high sampling DT-based NILM framework show: (a) a remarkably good performance of the model, despite the limited data, proving that the utilization of harmonics contributes to an improved classification, and (b) the applicability of the model to real-time applications given that the whole procedure from initial data processing to image classification (i.e., appliance identification) lasts less than 1 s.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 7","pages":""},"PeriodicalIF":3.4,"publicationDate":"2025-04-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10489-025-06512-3.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143793129","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

E-GAIL: efficient GAIL through including negative corruption and long-term rewards for robotic manipulations

IF 3.4 2区计算机科学

Applied Intelligence Pub Date : 2025-04-07 DOI: 10.1007/s10489-025-06335-2

Jiayi Tan, Gang Chen, Zeyuan Huang, Haofeng Liu, Marcelo H. Ang Jr

引用次数: 0

DCSR: A deep continual learning-based scheme for image super resolution using knowledge distillation

IF 3.4 2区计算机科学

Applied Intelligence Pub Date : 2025-04-05 DOI: 10.1007/s10489-025-06490-6

Alireza Esmaeilzehi, Hossein Zaredar, M. Omair Ahmad

{"title":"DCSR: A deep continual learning-based scheme for image super resolution using knowledge distillation","authors":"Alireza Esmaeilzehi, Hossein Zaredar, M. Omair Ahmad","doi":"10.1007/s10489-025-06490-6","DOIUrl":"10.1007/s10489-025-06490-6","url":null,"abstract":"<div><p>Deep neural networks have revolutionized the design of image super resolution schemes, in view of their capability of learning suitable features of high-resolution images. The performance of the deep image super resolution networks is very dependent on the distribution of the samples used for their training process. When the deep super resolution networks try to learn super-resolving the low-resolution images from different distributions in a sequential manner, they can only provide high performance for the most recent low-resolution image distribution used in their training process. In view of this, and in order to address the forgetting problem of the super resolution networks during learning from a new distribution of the low-resolution images, in this paper, we propose a continual learning-based scheme, which is developed based on the knowledge distillation technique. Specifically, our proposed deep continual learning-based image super resolution method aims at retaining the knowledge obtained from the previously learned distribution of the training samples, while learning the new distribution as efficiently as possible. To achieve this, the proposed scheme employs the supervision of the signals produced by multiple teacher networks. The results of the extensive experimentation show the effectiveness of the various ideas employed in the development of the proposed method. Further, it is shown that the proposed scheme outperforms the various state-of-the-art image super resolution methods when they are subjected to learning different distributions of the low-resolution images.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 7","pages":""},"PeriodicalIF":3.4,"publicationDate":"2025-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143777998","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Intention-aware neural networks with session disentanglement for noise filtering in session-based recommendation

IF 3.4 2区计算机科学

Applied Intelligence Pub Date : 2025-04-05 DOI: 10.1007/s10489-025-06519-w

Feihu Huang, Haoyu Xu, Ning Yang, Jince Wang, Peiyu Yi, Yuan Jiang

{"title":"Intention-aware neural networks with session disentanglement for noise filtering in session-based recommendation","authors":"Feihu Huang, Haoyu Xu, Ning Yang, Jince Wang, Peiyu Yi, Yuan Jiang","doi":"10.1007/s10489-025-06519-w","DOIUrl":"10.1007/s10489-025-06519-w","url":null,"abstract":"<div><p>Session-based recommendation is a significant and practical approach in predicting the next action of anonymous users within a recommendation system. However, accurate recommendations remain challenging due to limited information. Recently, many works based on neural networks have been proposed to address this task. Nevertheless, these works tend to focus solely on modeling item relationships while neglecting the importance of sessions and exhibiting suboptimal performance in handling noise items within current sessions. To address these issues, this paper proposes an Intention-Aware Neural Networks with Session Disentanglement (IANNSD) that incorporates session modeling and user intent as key factors. Specifically, in the local relationship encoder (LRE), we compute the similarity between the current session and its neighboring items to alleviate the impact of noise neighbor items on recommendation accuracy. In the global relationship encoder (GRE), sessions serve as a constraint for refining the intent distribution of each item, and a highway network is utilized to optimize the outputs of GRE. Additionally, we design a label optimization module to assist model training. Extensive experiments are carried out on three real datasets, and the experimental results demonstrate that IANNSD surpasses state-of-the-art models in session-based recommendation performance.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 7","pages":""},"PeriodicalIF":3.4,"publicationDate":"2025-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143777997","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Multi-perspective semantic decoupling and enhancement in graph attention network for knowledge graph completion 图注意网络中的多视角语义解耦和增强，用于完成知识图谱

IF 3.4 2区计算机科学

Applied Intelligence Pub Date : 2025-04-05 DOI: 10.1007/s10489-025-06520-3

Tianyi Xu, Yan Wang, Wenbin Zhang, Yue Zhao, Jian Yu, Mei Yu, Jiujiang Guo, Mankun Zhao

{"title":"Multi-perspective semantic decoupling and enhancement in graph attention network for knowledge graph completion","authors":"Tianyi Xu, Yan Wang, Wenbin Zhang, Yue Zhao, Jian Yu, Mei Yu, Jiujiang Guo, Mankun Zhao","doi":"10.1007/s10489-025-06520-3","DOIUrl":"10.1007/s10489-025-06520-3","url":null,"abstract":"<div><p>Knowledge Graphs (KGs) are semantic repositories that describe the real world and have been widely applied in various downstream applications. However, KGs still have many incomplete facts, so Knowledge Graph Completion (KGC) is proposed to infer missing facts. Among them, Graph Attention Network-based models (GATs) show great power. However, GATs have two flaws in handling multiple semantics of entities in relational context: (1) Current GATs fail to distinguish the various semantics of the entity which are exhibited by the relations from different perspectives. (2) Existing GATs cannot capture the similar semantics of different entities which are presented by the relations from the same perspective. Hence, we propose a graph attention network based on multi-perspective semantic decoupling and enhancement (MSDE). To capture diverse semantics in the relational context, we classify relations to partition entity multi-perspective semantics, and then we use graph attention networks to obtain multi-perspective decoupled embeddings of entities. To capture semantically similar entities, we select multi-perspective similar entities based on multi-perspective conditional entropy and high-order similar neighbors based on multi-perspective decoupled embedding. Finally, we use an attention decay network to aggregate multi-perspective similar entities and high-order similar neighbors to update entity feature embeddings. Experimental results show that MSDE exhibits marked performance gains compared to other state-of-the-art (sota) models. Significantly, the MRR indicator improves by 6.5% on the FB15K-237 dataset, by 2.3% on the WN18RR dataset, by 7.3% on the Kinship dataset and by 9.2% on the YAGO3-10 over the sota models.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 7","pages":""},"PeriodicalIF":3.4,"publicationDate":"2025-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143777996","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

How to find opinion leader on the online social network?

IF 3.4 2区计算机科学

Applied Intelligence Pub Date : 2025-04-04 DOI: 10.1007/s10489-025-06525-y

Bailu Jin, Mengbang Zou, Zhuangkun Wei, Weisi Guo

{"title":"How to find opinion leader on the online social network?","authors":"Bailu Jin, Mengbang Zou, Zhuangkun Wei, Weisi Guo","doi":"10.1007/s10489-025-06525-y","DOIUrl":"10.1007/s10489-025-06525-y","url":null,"abstract":"<div><p>Online social networks (OSNs) provide a platform for individuals to share information, exchange ideas, and build social connections beyond in-person interactions. For a specific topic or community, opinion leaders are individuals who have a significant influence on others’ opinions. Detecting opinion leaders and modeling influence dynamics is crucial as they play a vital role in shaping public opinion and driving conversations. Existing research have extensively explored various graph-based and psychology-based methods for detecting opinion leaders, but there is a lack of cross-disciplinary consensus between definitions and methods. For example, node centrality in graph theory does not necessarily align with the opinion leader concepts in social psychology. This review paper aims to address this multi-disciplinary research area by introducing and connecting the diverse methodologies for identifying influential nodes. The key novelty is to review connections and cross-compare different multi-disciplinary approaches that have origins in: social theory, graph theory, compressed sensing theory, and control theory. Our first contribution is to develop cross-disciplinary discussion on how they tell a different tale of networked influence. Our second contribution is to propose trans-disciplinary research method on embedding socio-physical influence models into graph signal analysis. We showcase inter- and trans-disciplinary methods through a Twitter case study to compare their performance and elucidate the research progression with relation to psychology theory. We hope the comparative analysis can inspire further research in this cross-disciplinary area.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 7","pages":""},"PeriodicalIF":3.4,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://link.springer.com/content/pdf/10.1007/s10489-025-06525-y.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143769838","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Exploring the application of ChatGPT in scientific topic analysis: a novel paradigm for enhanced analysis and efficiency

IF 3.4 2区计算机科学

Applied Intelligence Pub Date : 2025-04-04 DOI: 10.1007/s10489-025-06498-y

Muretijiang Muhetaer, Fan Hao

{"title":"Exploring the application of ChatGPT in scientific topic analysis: a novel paradigm for enhanced analysis and efficiency","authors":"Muretijiang Muhetaer, Fan Hao","doi":"10.1007/s10489-025-06498-y","DOIUrl":"10.1007/s10489-025-06498-y","url":null,"abstract":"<div><p>Latent Dirichlet Allocation (LDA) is a powerful text analysis tool that has been widely used in literature to reveal the development trends of disciplines and fields, thereby greatly broadening the frontier of text mining and knowledge discovery. However, as a probability model based on word frequency statistics, LDA has inherent limitations in its inability to deeply understand the deep meaning of words in a document set. Although some researchers have attempted to combine LDA with other deep learning models, such as BERT and BiLSTM, in order to improve the effectiveness of topic modeling, the progress achieved has not been significant. In this study, we innovatively propose to combine the text comprehension ability of ChatGPT with the statistical ability of LDA model, aiming to further improve the accuracy and depth of topic modeling. Specifically, we first conduct topic modeling on the target text using the LDA topic model to obtain a topic-word matrix. Then, we input the word set corresponding to each topic in the matrix into the ChatGPT model with an appropriate prompt template to obtain a topic name-description table that accurately describes the topic. Finally, we input the content of each target text and the corresponding topic name-description table into the ChatGPT model to obtain the topic classification result for each text. In addition, we also conduct quantitative evaluation on the proposed method through calculating similarity based on BERT's word embedding vector. The experimental results show that our proposed ChatGPT + LDA method can significantly enhance the effectiveness of topic modeling, bringing new breakthroughs to the field of text analysis and knowledge discovery.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 7","pages":""},"PeriodicalIF":3.4,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143778000","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Federated fine-grained prompts for vision-language models based on open-vocabulary object detection

IF 3.4 2区计算机科学

Applied Intelligence Pub Date : 2025-04-04 DOI: 10.1007/s10489-025-06527-w

Yu Li

{"title":"Federated fine-grained prompts for vision-language models based on open-vocabulary object detection","authors":"Yu Li","doi":"10.1007/s10489-025-06527-w","DOIUrl":"10.1007/s10489-025-06527-w","url":null,"abstract":"<div><p>Vision-language models can be used for open-vocabulary object detection. The existing methods suffer from low matching accuracy between prompt and image regions, as well as limited generalization capability as they adopt a data-centralized model training approach that ignores data heterogeneity. To alleviate these issues, we propose a federated fine-grained prompts learning method called FFPLearning, for open-vocabulary object detection using vision-language models. Specifically, FFPLearning quantifies the quality of proposals using pre-fused EoG (Energy of Gradient) and IoU (Intersection over Union) scores and organizes them into individual groups. Then learnable fine-grained prompts are trained to align the grouped region proposals in the feature space. A momentum update algorithm is designed to assess the quality of each participating client in the federated learning. Additionally, a Transformer-based feedback aggregation algorithm is designed to thoroughly leverage the semantic information from prompts and aggregate them based on the qualities of clients. Comprehensive evaluations on COCO and LVIS datasets demonstrate that FFPLearning is very effective, with +5.8 Novel AP50 and +3.3 APr improvements compared with existing state-of-the-art methods.</p></div>","PeriodicalId":8041,"journal":{"name":"Applied Intelligence","volume":"55 7","pages":""},"PeriodicalIF":3.4,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143777999","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0