Hyperspectral images have a wide range of applications in many fields. However, when hyperspectral images are captured by spectrometers, there is inevitably considerable noise, which affects subsequent research. In recent years, many hyperspectral image denoising methods based on low-rank representations have been proposed. Artificial intelligence denoising methods are also popular. However, the research on multi noise denoising is rarely mentioned, and most literatures only focus on one noise in hyperspectral images. Thus, we propose a denoising model for hyperspectral image based on dual graph-regularized low-rank representation, which can not only reduce multiple types of noise simultaneously, but also preserves details of the original image. In particular, this is the first time that the dual low-rank representation and dual graph regularizations are used on hyperspectral images. We solve this method using the linearized alternating direction method with adaptive penalty. Finally, we conduct experiments on simulated and real data sets to verify the effectiveness of our method. The experimental results show that our method can not only effectively remove a variety of mixed noises, but also well retain the details of the image.
{"title":"Dual graph-regularized low-rank representation for hyperspectral image denoising","authors":"Chengcai Leng , Mingpei Tang , Zhao Pei , Jinye Peng , Anup Basu","doi":"10.1016/j.engappai.2024.109659","DOIUrl":"10.1016/j.engappai.2024.109659","url":null,"abstract":"<div><div>Hyperspectral images have a wide range of applications in many fields. However, when hyperspectral images are captured by spectrometers, there is inevitably considerable noise, which affects subsequent research. In recent years, many hyperspectral image denoising methods based on low-rank representations have been proposed. Artificial intelligence denoising methods are also popular. However, the research on multi noise denoising is rarely mentioned, and most literatures only focus on one noise in hyperspectral images. Thus, we propose a denoising model for hyperspectral image based on dual graph-regularized low-rank representation, which can not only reduce multiple types of noise simultaneously, but also preserves details of the original image. In particular, this is the first time that the dual low-rank representation and dual graph regularizations are used on hyperspectral images. We solve this method using the linearized alternating direction method with adaptive penalty. Finally, we conduct experiments on simulated and real data sets to verify the effectiveness of our method. The experimental results show that our method can not only effectively remove a variety of mixed noises, but also well retain the details of the image.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109659"},"PeriodicalIF":7.5,"publicationDate":"2024-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720592","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}
The dynamic nature of interconnected data evolving over time poses significant challenges for graph representation and reasoning, particularly as temporal knowledge graphs scale in size and complexity. Existing models like TPComplEx (Time Perspective Complex Embedding) leverage tensor decomposition techniques to capture temporal dynamics, but their static weighting approach often lacks the flexibility needed to adapt to the nuanced evolution of relationships and entities. This rigidity can lead to missed temporal dependencies and loss of valuable insights, especially in large-scale graphs comprising millions or even billions of factual entries. To overcome these limitations, we propose FTPComplEx (Flexible Time Perspective Complex Embedding), a novel embedding model that introduces adjustable weights to dynamically modulate the influence of temporal information. This flexibility enables FTPComplEx to more accurately capture the intricate interactions between entities, relations, and time, providing a more robust understanding of temporal dynamics within knowledge graphs. Our extensive evaluations on benchmark datasets, including YAGO15k, ICEWS, and GDELT, demonstrate that FTPComplEx achieves state-of-the-art results, outperforming TPComplEx and other existing models. Notably, on the YAGO15k dataset, FTPComplEx achieves a 9.04% improvement in Mean Reciprocal Rank (MRR) and an 11.35% increase in Hits@1, demonstrating its effectiveness in managing complex temporal relationships. Further analysis shows that FTPComplEx maintains strong performance even with lower-rank embeddings, significantly reducing computational costs while maintaining accuracy.
{"title":"FTPComplEx: A flexible time perspective approach to temporal knowledge graph completion","authors":"Ngoc-Trung Nguyen , Thuc Ngo , Nguyen Hoang , Thanh Le","doi":"10.1016/j.engappai.2024.109717","DOIUrl":"10.1016/j.engappai.2024.109717","url":null,"abstract":"<div><div>The dynamic nature of interconnected data evolving over time poses significant challenges for graph representation and reasoning, particularly as temporal knowledge graphs scale in size and complexity. Existing models like TPComplEx (Time Perspective Complex Embedding) leverage tensor decomposition techniques to capture temporal dynamics, but their static weighting approach often lacks the flexibility needed to adapt to the nuanced evolution of relationships and entities. This rigidity can lead to missed temporal dependencies and loss of valuable insights, especially in large-scale graphs comprising millions or even billions of factual entries. To overcome these limitations, we propose FTPComplEx (Flexible Time Perspective Complex Embedding), a novel embedding model that introduces adjustable weights to dynamically modulate the influence of temporal information. This flexibility enables FTPComplEx to more accurately capture the intricate interactions between entities, relations, and time, providing a more robust understanding of temporal dynamics within knowledge graphs. Our extensive evaluations on benchmark datasets, including YAGO15k, ICEWS, and GDELT, demonstrate that FTPComplEx achieves state-of-the-art results, outperforming TPComplEx and other existing models. Notably, on the YAGO15k dataset, FTPComplEx achieves a 9.04% improvement in Mean Reciprocal Rank (MRR) and an 11.35% increase in Hits@1, demonstrating its effectiveness in managing complex temporal relationships. Further analysis shows that FTPComplEx maintains strong performance even with lower-rank embeddings, significantly reducing computational costs while maintaining accuracy.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109717"},"PeriodicalIF":7.5,"publicationDate":"2024-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720772","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}
Pub Date : 2024-11-23DOI: 10.1016/j.engappai.2024.109633
Muhammad Akram , Maheen Sultan , Cengiz Kahraman
The Chinese economy is one of the largest and most dynamic economies in the world. Over the past few decades, China has experienced rapid economic growth from agrarian to industrial powerhouse fueled by manufacturing, exports, and services. However, this rapid growth has also brought about challenges, including environmental issues like water contamination. The indulgence of cadmium metal in regular used water can cause serious health issues, including kidney damage and cancer. Many strategies have been implemented for treatment of water contamination. The main focus of this research is to introduce a novel methodology for treatment of cadmium contaminated water problem in China. This study seeks to demonstrate the multi-criteria group decision-making ability based on the outranking relations within the confines of a contemporary, well-organized and extremely flexible model of spherical fuzzy rough numbers. Spherical fuzzy rough numbers, amalgamation of rough numbers with traditional spherical fuzzy numbers, make the use of membership, non-membership and neutral membership degrees along with the manipulation of the subjectivity and reliance on objective uncertainties. The combination of spherical fuzzy rough numbers with an outranking multi-criteria group decision making technique, Elimination and Choice Expressing Reality, integrates spherical fuzzy logic to handle uncertainty and imprecision in multi-criteria decision-making. This approach captures degrees of uncertainty and hesitancy with spherical fuzzy numbers, improving the handling of imprecise information. The working mechanism involves generation of outranking relations among alternatives by comparing predominant and subdominant options, calculating score degrees, concordance and discordance sets, and incorporating subjective spherical fuzzy rough criteria weights. Unlike traditional methods that use crisp or conventional fuzzy numbers, this technique provides a more reliable and flexible evaluation by integrating rough set theory for better handling of imprecision and uncertainty. Finally, an outranking graph is drawn that points from the supreme option to inferior one. The legitimacy of the proposed technique is, then, testified by making its comparison with other existing techniques.
{"title":"An integrated outranking technique with spherical fuzzy rough numbers for the treatment of cadmium-contaminated water problem in China","authors":"Muhammad Akram , Maheen Sultan , Cengiz Kahraman","doi":"10.1016/j.engappai.2024.109633","DOIUrl":"10.1016/j.engappai.2024.109633","url":null,"abstract":"<div><div>The Chinese economy is one of the largest and most dynamic economies in the world. Over the past few decades, China has experienced rapid economic growth from agrarian to industrial powerhouse fueled by manufacturing, exports, and services. However, this rapid growth has also brought about challenges, including environmental issues like water contamination. The indulgence of cadmium metal in regular used water can cause serious health issues, including kidney damage and cancer. Many strategies have been implemented for treatment of water contamination. The main focus of this research is to introduce a novel methodology for treatment of cadmium contaminated water problem in China. This study seeks to demonstrate the multi-criteria group decision-making ability based on the outranking relations within the confines of a contemporary, well-organized and extremely flexible model of spherical fuzzy rough numbers. Spherical fuzzy rough numbers, amalgamation of rough numbers with traditional spherical fuzzy numbers, make the use of membership, non-membership and neutral membership degrees along with the manipulation of the subjectivity and reliance on objective uncertainties. The combination of spherical fuzzy rough numbers with an outranking multi-criteria group decision making technique, Elimination and Choice Expressing Reality, integrates spherical fuzzy logic to handle uncertainty and imprecision in multi-criteria decision-making. This approach captures degrees of uncertainty and hesitancy with spherical fuzzy numbers, improving the handling of imprecise information. The working mechanism involves generation of outranking relations among alternatives by comparing predominant and subdominant options, calculating score degrees, concordance and discordance sets, and incorporating subjective spherical fuzzy rough criteria weights. Unlike traditional methods that use crisp or conventional fuzzy numbers, this technique provides a more reliable and flexible evaluation by integrating rough set theory for better handling of imprecision and uncertainty. Finally, an outranking graph is drawn that points from the supreme option to inferior one. The legitimacy of the proposed technique is, then, testified by making its comparison with other existing techniques.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109633"},"PeriodicalIF":7.5,"publicationDate":"2024-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720591","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}
Pub Date : 2024-11-23DOI: 10.1016/j.engappai.2024.109677
Chunfeng Wan , Jiale Hou , Guangcai Zhang , Shuai Gao , Youliang Ding , Sugong Cao , Hao Hu , Songtao Xue
Machine learning algorithms can sensitively capture the characteristics of vortex induced vibration (VIV) of the girder in long span bridge from the extensive historical data accumulated by structural health monitoring (SHM) system over several years. These algorithms have gradually become a promising method of VIV identification. However, the algorithms proposed by previous researchers require historical VIV data to select the threshold or parameters to identify VIV. Most long-span bridges have not recorded a significant amount of VIV data since VIV is rare, or the bridge were not equipped with SHM system before. This study proposes an adaptive VIV identification method based on domain adaptation methods, which can identify VIV in real-time or in historical monitoring datasets of the target bridge without prior VIV information or parameter settings. The strong generalization ability of the proposed method is verified on the SHM dataset of two long-span suspension bridges in China. It is found that the VIV recognition accuracy of the balanced distribution adaptation (BDA) based VIV identification method is higher than that of other algorithms. In this study, the BDA based algorithm is also applied to the 8 months monitoring datasets of a long span bridge and successfully identifies more than 20 VIV events of the main girder, which has shown the stability and accuracy of the proposed algorithm.
{"title":"Domain adaptation based automatic identification method of vortex induced vibration of long-span bridges without prior information","authors":"Chunfeng Wan , Jiale Hou , Guangcai Zhang , Shuai Gao , Youliang Ding , Sugong Cao , Hao Hu , Songtao Xue","doi":"10.1016/j.engappai.2024.109677","DOIUrl":"10.1016/j.engappai.2024.109677","url":null,"abstract":"<div><div>Machine learning algorithms can sensitively capture the characteristics of vortex induced vibration (VIV) of the girder in long span bridge from the extensive historical data accumulated by structural health monitoring (SHM) system over several years. These algorithms have gradually become a promising method of VIV identification. However, the algorithms proposed by previous researchers require historical VIV data to select the threshold or parameters to identify VIV. Most long-span bridges have not recorded a significant amount of VIV data since VIV is rare, or the bridge were not equipped with SHM system before. This study proposes an adaptive VIV identification method based on domain adaptation methods, which can identify VIV in real-time or in historical monitoring datasets of the target bridge without prior VIV information or parameter settings. The strong generalization ability of the proposed method is verified on the SHM dataset of two long-span suspension bridges in China. It is found that the VIV recognition accuracy of the balanced distribution adaptation (BDA) based VIV identification method is higher than that of other algorithms. In this study, the BDA based algorithm is also applied to the 8 months monitoring datasets of a long span bridge and successfully identifies more than 20 VIV events of the main girder, which has shown the stability and accuracy of the proposed algorithm.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109677"},"PeriodicalIF":7.5,"publicationDate":"2024-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720892","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}
Pub Date : 2024-11-23DOI: 10.1016/j.engappai.2024.109701
Xiaoyu Zhang , Desheng He , Junjie Wang , Shengkun Wang , Meixiang Gu
Extensive damage to pile-supported structures, often caused by earthquake-induced lateral spreading, has been reported frequently in numerous major earthquakes. To mitigate such damage, accurate prediction of the seismic behavior of the soil-pile-superstructure system (SPSS) has been extensively studied through experimental and numerical simulations. However, these methods typically require substantial time and high cost, making them challenging to adapt in practical engineering scenarios. This study successfully applied machine learning (ML) techniques to predict the maximum seismic response of the SPSS, offering a more efficient and flexible solution for engineers. Six ML algorithms were used: decision tree (DT), k-nearest neighbor (KNN), extreme gradient boosting (XGB), random forest (RF), artificial neural network (ANN), and Gaussian process regression (GPR). A detailed evaluation of these algorithms has shown that ML models can effectively predict the maximum displacement of both pile and soil. Notably, XGB outperformed other methods in terms of accuracy, stability, and efficiency. Furthermore, the study indicates that the velocity-dependent ground motion parameter, root mean square velocity (vRMS), effectively represents the ground motion parameters for accurately predicting maximum pile-soil displacement. This study demonstrates the potential of ML in geotechnical earthquake engineering, establishing a basis for further applications and contributing to enhanced seismic design of pile-supported structures in liquefiable soils.
据报道，在多次大地震中，由地震引起的横向扩展通常会对桩支撑结构造成大面积破坏。为了减轻这种破坏，人们通过实验和数值模拟对土壤-桩-上部结构系统（SPSS）的地震行为进行了广泛研究。然而，这些方法通常需要大量的时间和高昂的成本，使其在实际工程应用中面临挑战。本研究成功应用了机器学习（ML）技术来预测 SPSS 的最大地震响应，为工程师提供了更高效、更灵活的解决方案。研究采用了六种 ML 算法：决策树 (DT)、k-近邻 (KNN)、极梯度提升 (XGB)、随机森林 (RF)、人工神经网络 (ANN) 和高斯过程回归 (GPR)。对这些算法的详细评估表明，ML 模型可以有效预测桩和土的最大位移。值得注意的是，XGB 在准确性、稳定性和效率方面都优于其他方法。此外，研究还表明，与速度相关的地面运动参数--均方根速度（vRMS）能有效代表地面运动参数，从而准确预测桩土的最大位移。这项研究证明了 ML 在岩土地震工程中的潜力，为进一步应用奠定了基础，并有助于加强可液化土中桩支撑结构的抗震设计。
{"title":"Machine learning for predicting maximum displacement in soil-pile-superstructure systems in laterally spreading ground","authors":"Xiaoyu Zhang , Desheng He , Junjie Wang , Shengkun Wang , Meixiang Gu","doi":"10.1016/j.engappai.2024.109701","DOIUrl":"10.1016/j.engappai.2024.109701","url":null,"abstract":"<div><div>Extensive damage to pile-supported structures, often caused by earthquake-induced lateral spreading, has been reported frequently in numerous major earthquakes. To mitigate such damage, accurate prediction of the seismic behavior of the soil-pile-superstructure system (SPSS) has been extensively studied through experimental and numerical simulations. However, these methods typically require substantial time and high cost, making them challenging to adapt in practical engineering scenarios. This study successfully applied machine learning (ML) techniques to predict the maximum seismic response of the SPSS, offering a more efficient and flexible solution for engineers. Six ML algorithms were used: decision tree (DT), k-nearest neighbor (KNN), extreme gradient boosting (XGB), random forest (RF), artificial neural network (ANN), and Gaussian process regression (GPR). A detailed evaluation of these algorithms has shown that ML models can effectively predict the maximum displacement of both pile and soil. Notably, XGB outperformed other methods in terms of accuracy, stability, and efficiency. Furthermore, the study indicates that the velocity-dependent ground motion parameter, root mean square velocity (<em>v</em><sub><em>RMS</em></sub>), effectively represents the ground motion parameters for accurately predicting maximum pile-soil displacement. This study demonstrates the potential of ML in geotechnical earthquake engineering, establishing a basis for further applications and contributing to enhanced seismic design of pile-supported structures in liquefiable soils.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109701"},"PeriodicalIF":7.5,"publicationDate":"2024-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720593","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}
Pub Date : 2024-11-23DOI: 10.1016/j.engappai.2024.109637
Jose Ponce, Alvin Barbier, Carlos E. Palau, Carlos Guardiola
The Real Driving Emission (RDE) test became a critical part of the process conducted by manufacturers to fulfill the approval procedure of every new vehicle model. This test measures the regulated emissions from a vehicle during a trip, which follows a specific set of operation requirements, aiming to assess the vehicle’s emission levels in real-world conditions. Additionally, In-Service Conformity (ISC) tests, which consist in performing an RDE trip, were also introduced to demonstrate vehicles emissions compliance over their lifespan. Considering that modern vehicles embed exhaust emission sensors and connectivity capabilities, it is believed that there is an opportunity for manufacturers to leverage the data generated by these vehicles to forecast the outcomes of an ISC test. However, as this study presents through the analysis of an extensive database of more than 600 trips from a mild-hybrid diesel vehicle, none of the real-world trips might comply with all the driving requirements of the RDE standard. Faced with this outcome, this work proposes the application of a Genetic Algorithm (GA) optimization to construct virtual RDE trips from real-driving data. In particular, the proposed methodology leverages such algorithm to combine real driving fragments from various trips in order to align with the main RDE trip requirements. The methodology focuses on vehicle, engine, and exhaust after-treatment variables, utilizing signal optimization connections to create a realistic analysis of vehicle pollutants. The research suggests that a combination of vehicle speed, coolant temperature, exhaust temperature, and Selective Catalytic Reduction (SCR) load leads to a significant number of RDE-compliant results under simplified legislative conditions, from which emissions profiles could be assessed. The proposed methodology details the development of an Adaptive Genetic Algorithm (AGA) and the data pipeline to create specific RDE trips, offering the capability to customize the desired Driving Cycles (DC).
{"title":"A novel approach in constructing virtual real driving emission trips through genetic algorithm optimization","authors":"Jose Ponce, Alvin Barbier, Carlos E. Palau, Carlos Guardiola","doi":"10.1016/j.engappai.2024.109637","DOIUrl":"10.1016/j.engappai.2024.109637","url":null,"abstract":"<div><div>The Real Driving Emission (RDE) test became a critical part of the process conducted by manufacturers to fulfill the approval procedure of every new vehicle model. This test measures the regulated emissions from a vehicle during a trip, which follows a specific set of operation requirements, aiming to assess the vehicle’s emission levels in real-world conditions. Additionally, In-Service Conformity (ISC) tests, which consist in performing an RDE trip, were also introduced to demonstrate vehicles emissions compliance over their lifespan. Considering that modern vehicles embed exhaust emission sensors and connectivity capabilities, it is believed that there is an opportunity for manufacturers to leverage the data generated by these vehicles to forecast the outcomes of an ISC test. However, as this study presents through the analysis of an extensive database of more than 600 trips from a mild-hybrid diesel vehicle, none of the real-world trips might comply with all the driving requirements of the RDE standard. Faced with this outcome, this work proposes the application of a Genetic Algorithm (GA) optimization to construct virtual RDE trips from real-driving data. In particular, the proposed methodology leverages such algorithm to combine real driving fragments from various trips in order to align with the main RDE trip requirements. The methodology focuses on vehicle, engine, and exhaust after-treatment variables, utilizing signal optimization connections to create a realistic analysis of vehicle pollutants. The research suggests that a combination of vehicle speed, coolant temperature, exhaust temperature, and Selective Catalytic Reduction (SCR) load leads to a significant number of RDE-compliant results under simplified legislative conditions, from which emissions profiles could be assessed. The proposed methodology details the development of an Adaptive Genetic Algorithm (AGA) and the data pipeline to create specific RDE trips, offering the capability to customize the desired Driving Cycles (DC).</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109637"},"PeriodicalIF":7.5,"publicationDate":"2024-11-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720590","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}
Pub Date : 2024-11-22DOI: 10.1016/j.engappai.2024.109668
Yunxin Xie , Liangyu Jin , Chenyang Zhu , Weibin Luo , Qian Wang
Recent advancements in Artificial Intelligence (AI), particularly deep learning, have significantly improved lithology identification in reservoir exploration by leveraging micrographic rock imagery. Deep neural networks excel in feature extraction, enhancing classification accuracy. However, these models are prone to domain shifts, which often degrade their performance in real-world applications. This paper proposes an unsupervised domain adaptation framework that integrates Fisher linear discriminant analysis and Online Hard Example Mining (OHEM) to mitigate domain shifts and improve classification, particularly in datasets with imbalanced classes. The model employs a -balanced global–local domain discriminator to align feature distributions between different domains and introduces focal loss with class-wise weighted factors for better handling of imbalanced data. Additionally, an adapted version of OHEM identifies difficult samples during training, allowing the model to concentrate on challenging cases. The proposed method is validated on micrographic rock imagery from the Tibet, Qinghai, and Xinjiang regions, achieving an average accuracy of 83.2%, which is 13.8% higher than ResNet50 and at least 1% superior to other domain adaptation models. This research highlights the potential of AI-driven solutions in geoscientific applications and provides a robust framework for unsupervised lithology classification.
{"title":"Enhanced cross-domain lithology classification in imbalanced datasets using an unsupervised domain Adversarial Network","authors":"Yunxin Xie , Liangyu Jin , Chenyang Zhu , Weibin Luo , Qian Wang","doi":"10.1016/j.engappai.2024.109668","DOIUrl":"10.1016/j.engappai.2024.109668","url":null,"abstract":"<div><div>Recent advancements in Artificial Intelligence (AI), particularly deep learning, have significantly improved lithology identification in reservoir exploration by leveraging micrographic rock imagery. Deep neural networks excel in feature extraction, enhancing classification accuracy. However, these models are prone to domain shifts, which often degrade their performance in real-world applications. This paper proposes an unsupervised domain adaptation framework that integrates Fisher linear discriminant analysis and Online Hard Example Mining (OHEM) to mitigate domain shifts and improve classification, particularly in datasets with imbalanced classes. The model employs a <span><math><mi>ω</mi></math></span>-balanced global–local domain discriminator to align feature distributions between different domains and introduces focal loss with class-wise weighted factors for better handling of imbalanced data. Additionally, an adapted version of OHEM identifies difficult samples during training, allowing the model to concentrate on challenging cases. The proposed method is validated on micrographic rock imagery from the Tibet, Qinghai, and Xinjiang regions, achieving an average accuracy of 83.2%, which is 13.8% higher than ResNet50 and at least 1% superior to other domain adaptation models. This research highlights the potential of AI-driven solutions in geoscientific applications and provides a robust framework for unsupervised lithology classification.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109668"},"PeriodicalIF":7.5,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720597","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}
Pub Date : 2024-11-22DOI: 10.1016/j.engappai.2024.109582
Zhichao Hu , Likun Liu , Lina Ma , Xiangzhan Yu
With the development of various artificial intelligence (AI)–based applications, detecting anomalies and analyzing the root causes from massive data are critical to increasing the usability of AI. Fast, accurate root-cause analysis (RCA) that finds the main reason for an anomaly, as well as reasonable explanations, helps in solving problems effectively. Thus, RCA plays an important role in troubleshooting and fault diagnosis, making its application in data analysis crucial. Previous root-cause-localization approaches for multidimensional anomaly events encompass various techniques to reduce search space and have improved the localization performance. However, they do not effectively balance the requirements in terms of performance, compatibility, and interpretability. To solve these problems, we propose a new root-cause-localization method called SinkFlow. It provides a unified framework event-aggregation Graph (EAG) to describe the constraints of event aggregation and relations between events, so it can be easily generalized to various domains. SinkFlow introduces an applicable measure evaluation method for both fundamental and derived measures to quantify the impact of events. Also, it utilizes an optimal search strategy to reduce the search space based on the anomaly behavioral consistency and deviation significance. Our experimental results on semisynthetic datasets show that SinkFlow achieved better performance than other baselines and ran much faster, achieving a 1.88% increase of the F1-score and only 25% of the time cost of the second best localization method. In addition, SinkFlow offered clear, visible explanations of the localization results to answer the questions of why they are root causes and how the anomaly is formed.
{"title":"SinkFlow: Fast and traceable root-cause localization for multidimensional anomaly events","authors":"Zhichao Hu , Likun Liu , Lina Ma , Xiangzhan Yu","doi":"10.1016/j.engappai.2024.109582","DOIUrl":"10.1016/j.engappai.2024.109582","url":null,"abstract":"<div><div>With the development of various artificial intelligence (AI)–based applications, detecting anomalies and analyzing the root causes from massive data are critical to increasing the usability of AI. Fast, accurate root-cause analysis (RCA) that finds the main reason for an anomaly, as well as reasonable explanations, helps in solving problems effectively. Thus, RCA plays an important role in troubleshooting and fault diagnosis, making its application in data analysis crucial. Previous root-cause-localization approaches for multidimensional anomaly events encompass various techniques to reduce search space and have improved the localization performance. However, they do not effectively balance the requirements in terms of performance, compatibility, and interpretability. To solve these problems, we propose a new root-cause-localization method called <em>SinkFlow</em>. It provides a unified framework event-aggregation Graph (EAG) to describe the constraints of event aggregation and relations between events, so it can be easily generalized to various domains. <em>SinkFlow</em> introduces an applicable measure evaluation method for both fundamental and derived measures to quantify the impact of events. Also, it utilizes an optimal search strategy to reduce the search space based on the anomaly behavioral consistency and deviation significance. Our experimental results on semisynthetic datasets show that <em>SinkFlow</em> achieved better performance than other baselines and ran much faster, achieving a 1.88% increase of the F1-score and only 25% of the time cost of the second best localization method. In addition, <em>SinkFlow</em> offered clear, visible explanations of the localization results to answer the questions of why they are root causes and how the anomaly is formed.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109582"},"PeriodicalIF":7.5,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720680","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}
Pub Date : 2024-11-22DOI: 10.1016/j.engappai.2024.109607
Shahzaib Ashraf , Muhammad Naeem , Wania Iqbal , Hafiz Muhammad Athar Farid , Hafiz Muhammad Shakeel , Vladimir Simic , Erfan Babaee Tırkolaee
The emergence of the Internet of Things (IoT) for monitoring in real-time is geared towards sustainable energy consumption practices by taking control over energy loss. The promising potential of current IoT real-time monitoring systems paves the way for future developments in monitoring devices with eco-friendly sensing capabilities. As a result, the creation of effective IoT real-time monitoring devices targeted at decreasing energy loss becomes crucial. This modeling procedure falls under the realm of multiple-attribute group decision-making (MAGDM), aiming to integrate the Schweizer–Sklar (SS) -norm and -conorm within the disc spherical fuzzy (D-SF) framework. The objective is to enhance the flexibility of D-SF in dealing with intricate and uncertain data. The core focus of this research is on deriving SS -norm and -conorm for D-SF data, consequently introducing innovative aggregation operators. The article offers the fundamental D-SF operations using SS aggregation operators in a systematic manner, with thorough theorem justifications. A new MAGDM tool is presented, created simply to manage ambiguous and imprecise data utilizing the suggested operators. Our model is specifically designed to tackle the critical issue of reducing energy loss in IoT real-time monitoring systems. The research not only focuses on model accuracy but also emphasizes its effectiveness in solving this pressing problem, demonstrating significant advancements in sustainable energy practices. Moreover, the proposed aggregation operators are subjected to a comparative analysis. This comprehensive comparison not only enhances the operators’ efficacy but also underscores their relevance in real-world decision-making scenarios.
{"title":"Selection of Internet of Things-enabled sustainable real-time monitoring strategies for manufacturing processes using a disc spherical fuzzy Schweizer–Sklar aggregation model","authors":"Shahzaib Ashraf , Muhammad Naeem , Wania Iqbal , Hafiz Muhammad Athar Farid , Hafiz Muhammad Shakeel , Vladimir Simic , Erfan Babaee Tırkolaee","doi":"10.1016/j.engappai.2024.109607","DOIUrl":"10.1016/j.engappai.2024.109607","url":null,"abstract":"<div><div>The emergence of the Internet of Things (IoT) for monitoring in real-time is geared towards sustainable energy consumption practices by taking control over energy loss. The promising potential of current IoT real-time monitoring systems paves the way for future developments in monitoring devices with eco-friendly sensing capabilities. As a result, the creation of effective IoT real-time monitoring devices targeted at decreasing energy loss becomes crucial. This modeling procedure falls under the realm of multiple-attribute group decision-making (MAGDM), aiming to integrate the Schweizer–Sklar (SS) <span><math><mi>τ</mi></math></span>-norm and <span><math><mi>τ</mi></math></span>-conorm within the disc spherical fuzzy (D-SF) framework. The objective is to enhance the flexibility of D-SF in dealing with intricate and uncertain data. The core focus of this research is on deriving SS <span><math><mi>τ</mi></math></span>-norm and <span><math><mi>τ</mi></math></span>-conorm for D-SF data, consequently introducing innovative aggregation operators. The article offers the fundamental D-SF operations using SS aggregation operators in a systematic manner, with thorough theorem justifications. A new MAGDM tool is presented, created simply to manage ambiguous and imprecise data utilizing the suggested operators. Our model is specifically designed to tackle the critical issue of reducing energy loss in IoT real-time monitoring systems. The research not only focuses on model accuracy but also emphasizes its effectiveness in solving this pressing problem, demonstrating significant advancements in sustainable energy practices. Moreover, the proposed aggregation operators are subjected to a comparative analysis. This comprehensive comparison not only enhances the operators’ efficacy but also underscores their relevance in real-world decision-making scenarios.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109607"},"PeriodicalIF":7.5,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720675","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}
Pub Date : 2024-11-22DOI: 10.1016/j.engappai.2024.109644
Zeqi Wei , Hui Wang , Zhibin Zhao , Zheng Zhou , Ruqiang Yan
Gearbox fault diagnosis is crucial for the safe operation of mechanical systems. While Deep Learning (DL) has demonstrated promising results in this area, most existing methods rely on static supervised learning, lacking the dynamic, interactive learning capabilities similar to human decision-making. To tackle this issue, this study presents a novel approach that combines the strengths of Deep Reinforcement Learning (DRL) with the interpretability of a temporal shrinkage interpretable network. DRL integrates the perception abilities of DL with the decision-making capabilities of Reinforcement Learning (RL), offering a more comprehensive solution for complex challenges. In this method, gearbox fault diagnosis is formulated as a sequential decision problem within a Classification Markov Decision Process (CMDP). A multi-scale temporal shrinkage module is utilized to construct an interpretable network, which enhances model interpretability and reduces the negative impact of noisy data in harsh working conditions. The diagnosis agent autonomously learns the optimal classification policy, reducing the need for manual intervention and human expertise. Experimental results show excellent generalization and stability, achieving over 98.5% accuracy even in noisy conditions. It outperforms existing methods and highlights its robustness in challenging operational environments.
{"title":"Gearbox fault diagnosis based on temporal shrinkage interpretable deep reinforcement learning under strong noise","authors":"Zeqi Wei , Hui Wang , Zhibin Zhao , Zheng Zhou , Ruqiang Yan","doi":"10.1016/j.engappai.2024.109644","DOIUrl":"10.1016/j.engappai.2024.109644","url":null,"abstract":"<div><div>Gearbox fault diagnosis is crucial for the safe operation of mechanical systems. While Deep Learning (DL) has demonstrated promising results in this area, most existing methods rely on static supervised learning, lacking the dynamic, interactive learning capabilities similar to human decision-making. To tackle this issue, this study presents a novel approach that combines the strengths of Deep Reinforcement Learning (DRL) with the interpretability of a temporal shrinkage interpretable network. DRL integrates the perception abilities of DL with the decision-making capabilities of Reinforcement Learning (RL), offering a more comprehensive solution for complex challenges. In this method, gearbox fault diagnosis is formulated as a sequential decision problem within a Classification Markov Decision Process (CMDP). A multi-scale temporal shrinkage module is utilized to construct an interpretable network, which enhances model interpretability and reduces the negative impact of noisy data in harsh working conditions. The diagnosis agent autonomously learns the optimal classification policy, reducing the need for manual intervention and human expertise. Experimental results show excellent generalization and stability, achieving over 98.5% accuracy even in noisy conditions. It outperforms existing methods and highlights its robustness in challenging operational environments.</div></div>","PeriodicalId":50523,"journal":{"name":"Engineering Applications of Artificial Intelligence","volume":"139 ","pages":"Article 109644"},"PeriodicalIF":7.5,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142720888","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}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号