Article
Computer Science, Artificial Intelligence
Yongjiao Sun, Xin Yao, Xin Bi, Xuechun Huang, Xiangguo Zhao, Baiyou Qiao
Summary: Sea surface temperature (SST) is a crucial indicator for measuring sea heat and affecting human activities. A new method using graph learning to predict SST is proposed, which proves to be more efficient and accurate in dealing with data containing time-series information.
Article
Computer Science, Information Systems
Junran Wu, Ke Xu, Xueyuan Chen, Shangzhe Li, Jichang Zhao
Summary: This study proposes a novel framework to address the issues of long-term dependencies and chaotic properties in stock prediction. By transforming time series into complex networks and extracting structural information from the mapped graphs, the performance of the prediction model is improved. The effectiveness of the framework is validated through real-world stock data and trading simulations.
INFORMATION SCIENCES
(2022)
Article
Mathematics
Cheng Zhao, Ping Hu, Xiaohui Liu, Xuefeng Lan, Haiming Zhang
Summary: The ability to predict stock prices is essential for investment decisions, but the complexity of factors influencing stock prices has been extensively studied. Traditional methods that focus on time-series information for a single stock lack a holistic perspective. A time series relational model (TSRM) is proposed in this paper to integrate time and relationship information. The TSRM utilizes transaction data, K-means model, LSTM, and GCN to predict stock prices, yielding significant improvements in cumulative returns and maximum drawdown in the Chinese stock markets.
Article
Computer Science, Artificial Intelligence
Yusheng Huang, Xiaoyan Mao, Yong Deng
Summary: The degree sequence of the NVG transformation provides useful motif information for practical usage, as shown in a study on stock trend prediction. The proposed natural visibility encoding and moving window strategy have been proven effective and robust in classifying time series. Further investigation into the degree sequence of the NVG transformation is encouraged based on the success of the proposed framework.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Engineering, Civil
Zhiqiang Lv, Zesheng Cheng, Jianbo Li, Zhihao Xu, Zheng Yang
Summary: The complexity and spatial-temporal correlations in traffic scenarios pose challenges for traffic prediction research. Existing methods lack consideration of directional and hierarchical features among traffic nodes. This study proposes Tree Convolutional Network (TreeCN), a tree-based structure, to capture these features. Experimental results show that TreeCN performs well in both random uniform distribution scenarios and more complex small-scale aggregation scenarios, making it a promising method for handling complex traffic scenarios and improving prediction accuracy.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Mathematics
Lyuchao Liao, Zhiyuan Hu, Chih-Yu Hsu, Jinya Su
Summary: A novel robust Fourier Graph Convolution Network model is proposed for effective spatio-temporal pattern recognition of traffic flow data. The model includes a Fourier Embedding module and a stackable Spatial-Temporal ChebyNet layer. The Fourier Embedding module captures periodicity features based on Fourier series theory, while the Spatial-Temporal ChebyNet layer models the volatility features of traffic flow to improve system robustness.
Article
Computer Science, Artificial Intelligence
Atul Kumar Dwivedi, Umadevi Kaliyaperumal Subramanian, Jinsa Kuruvilla, Aby Thomas, D. Shanthi, Anandakumar Haldorai
Summary: Time-series prediction is a popular research topic with various applications. Telemetry data prediction is crucial for networking and information center control software. The concept of intuitionistic fuzzified time series is introduced to deal with non-determinism in time-series prediction.
Review
Engineering, Mechanical
Tao Wen, Huiling Chen, Kang Hao Cheong
Summary: The analysis of time series and images is significant in various fields. Visibility graph algorithms are used to map time series and images into different types of complex networks in order to explore their topological structure and information. By using local random walk algorithms and information fusion methods, time series can be forecasted, and images can be classified using machine learning models. The visibility graph algorithm outperforms existing algorithms in time series prediction and image classification, making complex networks an important tool for understanding the characteristics of time series and images.
NONLINEAR DYNAMICS
(2022)
Article
Optics
S. Gandhimathi Alias Usha, S. Vasuki
Summary: The paper proposed a method for time series analysis of multispectral images using hybrid graph cut segmentation and game theory classifier. The approach outperformed traditional algorithms in terms of stability and performance, demonstrating superiority in monitoring changes in the earth's surface information.
Article
Physics, Fluids & Plasmas
Sangwon Lee, Vipul Periwal, Junghyo Jo
Summary: Inferring dynamics from incomplete time series data is challenging, but an expectation maximization algorithm proposed in this study demonstrates effectiveness in restoring missing data points and inferring underlying network models. Balancing consistency between observed and missing data points is crucial for accurate model inference during iterative processes.
Article
Biochemical Research Methods
Shota Teramoto, Yusaku Uga
Summary: In this study, we developed a semi-automatic workflow that tracks individual root growth by vectorizing root system architecture (RSA) from time-series 3D images. This workflow can be applied to the time-series analysis of RSA development and plasticity.
Article
Computer Science, Artificial Intelligence
Xun Shi, Kuangrong Hao, Lei Chen, Bing Wei, Xiaoyan Liu
Summary: The paper introduces an improved graph convolution filter and a simple yet effective method for learning graph structure information, constructing a framework for multivariate time series prediction, with experimental results demonstrating the effectiveness of the model.
ADVANCED ENGINEERING INFORMATICS
(2022)
Article
Green & Sustainable Science & Technology
Hima Shaji, Lelitha Vanajakshi, Arun Tangirala
Summary: The accuracy of predicting bus travel time is an important step in improving the quality of public transportation. Previous studies have used chronological factors to identify significant regressors for predicting travel times. However, travel time patterns can vary depending on time and location. This study systematically analyzes the impact of different ways of presenting input data to the prediction algorithm. The results show that grouping the dataset and training separate models on them, particularly considering data-derived clusters, can significantly improve prediction accuracy.
Article
Mathematics, Applied
Felix Koester, Dhruvit Patel, Alexander Wikner, Lina Jaurigue, Kathy Luedge
Summary: We propose a new approach to dynamical system forecasting called data-informed-reservoir computing (DI-RC) that, while solely being based on data, yields increased accuracy, reduced computational cost, and mitigates tedious hyper-parameter optimization of the reservoir computer (RC).
Article
Mathematics
Seonghun Kim, Seockhun Bae, Yinhua Piao, Kyuri Jo
Summary: Integrating gene expression data and biological networks into the analysis framework for drug response prediction can improve prediction accuracy. DrugGCN successfully achieves this goal through graph convolutional network technology and demonstrates its success in biological data.
Article
Computer Science, Information Systems
Miltiadis Siavvas, Dimitrios Tsoukalas, Marija Jankovic, Dionysios Kehagias, Dimitrios Tzovaras
Summary: Vulnerability prediction is important for developing secure software by identifying and mitigating security risks early. The study suggests that technical debt indicators may have potential as security indicators, at both project and class levels.
ENTERPRISE INFORMATION SYSTEMS
(2022)
Article
Computer Science, Cybernetics
Sofia Segkouli, Dimitrios Giakoumis, Konstantinos Votis, Andreas Triantafyllidis, Ioannis Paliokas, Dimitrios Tzovaras
Summary: This study emphasizes the potential of personalized ICT solutions in improving the workability and well-being of older workers, and the need to present the risks from an ethical standpoint. The SmartFrameWorK ethics framework is proposed to address the uncertainties of workplace digitization and promote trust in digital technologies through its five-dimensional approach. Risk analysis highlights the importance of ethically aware practices for monitoring older workers' activities and behavior. Applying the SmartFrameWorK methodology in a case study provides evidence that personalized digital services can foster trust in users.
UNIVERSAL ACCESS IN THE INFORMATION SOCIETY
(2023)
Article
Computer Science, Cybernetics
Georgia Peleka, Dimitrios Sarafis, Ioannis Mariolis, Dimitrios Tzovaras
Summary: In this paper, a framework for generating photorealistic synthetic datasets using HDRI is proposed, containing all kinds of information required by computer vision algorithms. Furthermore, the framework demonstrates cross-domain knowledge transfer in a semantic segmentation scenario. It is found that deep neural networks trained with synthetic images or a combination of real and synthetic images perform equally or even better than those trained only on real images. This is the first work to successfully transfer knowledge from the synthetic domain to the real world using HDRI.
CYBERNETICS AND SYSTEMS
(2023)
Article
Materials Science, Multidisciplinary
Paschalis Charalampous, Nikolaos Kladovasilakis, Ioannis Kostavelis, Konstantinos Tsongas, Dimitrios Tzetzis, Dimitrios Tzovaras
Summary: This study aims to improve the mechanical properties of 3D printed parts in fused filament fabrication (FFF) process by using machine learning-based regression models and optimization techniques.
JOURNAL OF MATERIALS ENGINEERING AND PERFORMANCE
(2022)
Article
Computer Science, Interdisciplinary Applications
Thanasis Vafeiadis, Nikolaos Kolokas, Nikolaos Dimitriou, Angeliki Zacharaki, Murat Yildirim, Habibe Gulben Selvi, Dimosthenis Ioannidis, Dimitrios Tzovaras
Summary: This study presents a comparative assessment of 2DCNN and boosting methods for textile whiteness estimation, with WERegNet architecture outperforming ColorNet and XGBoost in terms of performance while being comparable to Random Forest.
SIMULATION MODELLING PRACTICE AND THEORY
(2022)
Article
Computer Science, Cybernetics
Andreas Triantafyllidis, Anastasios Alexiadis, Dimosthenis Elmas, Georgios Gerovasilis, Konstantinos Votis, Dimitrios Tzovaras
Summary: Childhood obesity is a significant public health challenge that is aggravated by lifestyle changes caused by the COVID-19 pandemic. A social robot-based platform shows potential in addressing childhood obesity by implementing an interactive motivational strategy and collecting behavioral data. Preliminary results from an experimental study demonstrate children's acceptance of the platform and its effectiveness in achieving recommended health goals.
UNIVERSAL ACCESS IN THE INFORMATION SOCIETY
(2023)
Article
Chemistry, Multidisciplinary
Asimina Dimara, Vasileios-Georgios Vasilopoulos, Alexios Papaioannou, Sotirios Angelis, Konstantinos Kotis, Christos-Nikolaos Anagnostopoulos, Stelios Krinidis, Dimosthenis Ioannidis, Dimitrios Tzovaras
Summary: This paper proposes an advanced IoT network management approach that supports the interoperability of multiple smart edge devices in the smart home network. It also applies IoT health-monitoring algorithms to detect and fix network anomalies, improving system stability and accuracy.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Information Systems
Athanasios Salamanis, Georgia Xanthopoulou, Dionysios Kehagias, Dimitrios Tzovaras
Summary: This paper introduces a deep learning-based approach for long-term tourism demand forecasting, which incorporates exogenous variables to improve forecasting accuracy. The proposed models were evaluated on real data from three hotels in Greece, demonstrating superior forecasting performance.
Article
Computer Science, Theory & Methods
Andrea Cimmino, Juan Cano-Benito, Alba Fernandez-Izquierdo, Christos Patsonakis, Apostolos C. Tsolakis, Raul Garcia-Castro, Dimosthenis Ioannidis, Dimitrios Tzovaras
Summary: Demand Response (DR) is essential for the EU energy markets, and standardising demand response data models has been a significant effort. However, existing proposals are usually centralised and hindered by heterogeneous data formats and models. This article presents a tool called CIM that allows DR systems to decentralise their architectures and exchange data transparently, even with systems following different DR standards. CIM provides a solid security framework and semantic interoperability layer for data exchange.
FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE
(2023)
Article
Automation & Control Systems
Apostolos Evangelidis, Nikolaos Dimitriou, Lampros Leontaris, Dimosthenis Ioannidis, Gregory Tinker, Dimitrios Tzovaras
Summary: Deep learning has made significant progress in industrial inspection and combining it with metrology has yielded impressive results. However, deploying metrology sensors in factories is challenging due to cost and special acquisition conditions. This article proposes a methodology that replaces a high-end sensor with a low-cost data-driven soft sensor model. It introduces a residual architecture (R(2)esNet) for quality inspection and an error-correction scheme to reduce noise impact. The methodology is evaluated in PCB manufacturing and achieves promising results, significantly reducing the inspection time compared to other methods.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Proceedings Paper
Computer Science, Information Systems
Nikolaos Siopis, Anastasios Alexiadis, Georgios Gerovasilis, Andreas Triantafyllidis, Konstantinos Votis, Dimitrios Tzovaras
Summary: HSmartBPM is an intelligent system tailored for the management of hypertension, providing personalized approaches for high blood pressure patients and healthcare professionals through components such as virtual agents, decision support systems, and shared care plan activities.
2022 IEEE-EMBS INTERNATIONAL CONFERENCE ON BIOMEDICAL AND HEALTH INFORMATICS (BHI) JOINTLY ORGANISED WITH THE IEEE-EMBS INTERNATIONAL CONFERENCE ON WEARABLE AND IMPLANTABLE BODY SENSOR NETWORKS (BSN'22)
(2022)
Proceedings Paper
Automation & Control Systems
George Margetis, Konstantinos C. Apostolakis, Nikolaos Dimitriou, Dimitrios Tzovaras, Constantine Stephanidis
Summary: This paper presents the OPTIMAI project architecture for zero-defect manufacturing (ZDM) applicable to various industrial sectors. By drawing parallels between the presented framework and two leading reference architectures, the authors provide a standards-based approach. The study examines cutting-edge technologies like blockchain, AI, and AR as both solutions for I4.0 and Industrial Internet of Things systems.
2022 IEEE 27TH INTERNATIONAL CONFERENCE ON EMERGING TECHNOLOGIES AND FACTORY AUTOMATION (ETFA)
(2022)
Proceedings Paper
Archaeology
Christina Tsita, Charalabos Georgiadis, Maria Zampeti, Evi Papavergou, Syrago Tsiara, Alexandros Pedefoudas, Dionysios Kehagias
Summary: The number of exhibitions in virtual museums has been increasing in recent years, as cultural heritage institutions strive to communicate with their audiences. Compared to using panoramic photos and virtual exhibitions, virtual reality museums offer more complex and educational hands-on activities. This paper presents a theoretical and practical approach to understanding contemporary art through virtual reality museums, enhancing the educational value of the experience.
TRANDISCIPLINARY MULTISPECTRAL MODELLING AND COOPERATION FOR THE PRESERVATION OF CULTURAL HERITAGE, TMM_CH 2021
(2022)
Review
Computer Science, Information Systems
Sofia Polymeni, Evangelos Athanasakis, Georgios Spanos, Konstantinos Votis, Dimitrios Tzovaras
Summary: Climate change has drawn attention to the research community in the natural environment sector in recent years, with the advent of IoT and AI technologies providing new opportunities and methods for environmental research. Through a systematic literature review of recent studies, it was found that many IoT-based prediction models have been applied to address various environmental issues in the past few years, with promising results in the majority of cases.
INTERNET OF THINGS
(2022)
Proceedings Paper
Computer Science, Interdisciplinary Applications
Chrysoula Oikonomou, Charalampos S. Kouzinopoulos, Dimosthenis Ioannidis, Dimitrios Tzovaras
Summary: This paper presents a lightweight scheme for Data in Transit Encryption (DiTE) for embedded devices. The scheme uses dynamic keys generated based on communication channel characteristics, providing enhanced security level.
2022 11TH MEDITERRANEAN CONFERENCE ON EMBEDDED COMPUTING (MECO)
(2022)