Article
Computer Science, Artificial Intelligence
Iurii Konovalenko, Andre Ludwig
Summary: Real-time temperature monitoring is crucial in cold pharmaceutical supply chains to prevent product quality deterioration from extreme temperature exposure. A new hybrid k-NN algorithm, based on principles of local similarity and neighborhood homogeneity, outperforms traditional k-NN in accuracy and precision for temperature alarms in pharmaceutical supply chains.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Automation & Control Systems
J. A. Romero-del-Castillo, Manuel Mendoza-Hurtado, Domingo Ortiz-Boyer, Nicolas Garcia-Pedrajas
Summary: Multi-label learning is an important field in machine learning research, and the multi-label k-nearest neighbor method is one of the most successful algorithms. However, allocating the appropriate value of k is a challenge in difficult classification tasks, as different regions may require different k values. We propose a simple yet powerful method to set local k values, obtaining the optimal value by optimizing the local effect of different k values near each prototype.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2022)
Article
Computer Science, Artificial Intelligence
Maximiliano Cubillos, Sanne Wohlk, Jesper N. Wulff
Summary: This study proposes a bi-objective algorithm based on the k-nearest neighbors method for imputing missing values in data with continuous variables and multilevel structures. Results from simulation studies show that the proposed method outperforms benchmark methods in cases with high intraclass correlation, reducing estimation bias and coefficient of determination.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Engineering, Chemical
Jian Wang, Zhe Zhou, Zuxin Li, Shuxin Du
Summary: This paper proposes a new fault detection scheme using the mutual k-nearest neighbor (MkNN) method, which improves the performance of fault detection by eliminating the influence of outliers and directly detecting fault samples without MkNNs.
Article
Computer Science, Information Systems
Mahesh Thyluru Ramakrishna, Vinoth Kumar Venkatesan, Rajat Bhardwaj, Surbhi Bhatia, Mohammad Khalid Imam Rahmani, Saima Anwar Lashari, Aliaa M. Alabdali
Summary: The emergence of social media platforms has greatly improved social connections. However, finding the right friends remains a challenge. This study proposes a social and semantic-based collaborative filtering approach to enhance personalized recommendations. The results show that this approach improves recommendation accuracy and addresses the issues of sparsity and cold start.
Article
Automation & Control Systems
Zekang Bian, Chi Man Vong, Pak Kin Wong, Shitong Wang
Summary: This study proposes a novel classification method based on FKNN called A-FKNN that learns the optimal k value for each testing sample, and a faster version called FA-FKNN is designed. Experimental results show that both A-FKNN and FA-FKNN outperform other methods in terms of classification accuracy, with FA-FKNN having a shorter running time.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Computer Science, Artificial Intelligence
Danny Hartanto Djarum, Zainal Ahmad, Jie Zhang
Summary: RBOSR is a new approach that improves the performance and efficiency of the PM2.5 stacked model. It significantly reduces training time and outperforms the original model.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Yongda Cai, Joshua Zhexue Huang, Jianfei Yin
Summary: This paper proposes a new method called adaptive k-nearest neighbors similarity graph (AKNNG) for constructing a better graph structure. By assigning different k values to different data points and automatically adjusting the k value based on the similarity graph, the AKNNG method improves clustering accuracies and reduces construction time.
Article
Computer Science, Artificial Intelligence
Zhibin Pan, Yiwei Pan, Yidi Wang, Wei Wang
Summary: The LMKNN classifier has better performance and robustness compared to the KNN classifier, but the unreliable nearest neighbor selection rule and single local mean vector strategy severely impact its classification performance.
Article
Computer Science, Information Systems
Shanshan Liu, Pedro Reviriego, Jose Alberto Hernandez, Fabrizio Lombardi
Summary: This paper explores how to provide protection and error tolerance for classifiers by exploiting the algorithmic properties, applied to the k Nearest Neighbors classifier, and proposes a time-based modular redundancy scheme to reduce the number of re-computations needed effectively.
IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTING
(2021)
Article
Computer Science, Hardware & Architecture
Payam Bahrani, Behrouz Minaei-Bidgoli, Hamid Parvin, Mitra Mirzarezaee, Ahmad Keshavarz
Summary: Despite advancements in recommender systems, there is still room for improvement. This study developed an improved method using weighted averaging and a Gaussian mixture model, which showed more accurate results and faster execution time compared to traditional methods.
JOURNAL OF SUPERCOMPUTING
(2023)
Article
Computer Science, Information Systems
Jiawei Yang, Yu Chen, Sylwan Rahardja
Summary: Traditional outlier detectors have neglected the group-level factor in calculating outlier scores for objects in data, resulting in the inability to capture collective outliers. To address this issue, a framework called neighborhood representative (NR) is proposed, enabling existing outlier detectors to efficiently detect outliers, including collective outliers, while maintaining computational integrity. By selecting representative objects, scoring them, and applying the score to collective objects, NR achieves this without altering existing detectors. NR is compatible with existing detectors and improves performance on eleven real-world datasets by an average of 8% (0.72 to 0.78 AUC) relative to twelve state-of-the-art outlier detectors. The implementation of NR can be found at www.OutlierNet.com for reproducibility. (c) 2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
INFORMATION SCIENCES
(2023)
Article
Statistics & Probability
Emre Demirkaya, Yingying Fan, Lan Gao, Jinchi Lv, Patrick Vossler, Jingbo Wang
Summary: This work introduces a novel two-scale DNN method by linearly combining two DNN estimators with different subsampling scales to reduce bias and achieve the optimal nonparametric convergence rate under the fourth-order smoothness condition.
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
(2022)
Article
Computer Science, Artificial Intelligence
Zhiyong Li, Tao Li, Junjiang He, Yongbin Zhu, Yunpeng Wang
Summary: A new negative selection algorithm incorporating the k-NN algorithm is proposed in this study to improve abnormality detection rate by classifying instances using detectors and neighboring instances. Both theoretical analysis and experimental results demonstrate that the algorithm has a higher detection rate in most cases.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Information Systems
Junnan Li, Qingsheng Zhu, Quanwang Wu, Zhu Fan
Summary: Class imbalance is a significant factor leading to performance deterioration in classifiers. Techniques such as SMOTE and its extension, NaNSMOTE, have been successful in addressing this issue and have been proven effective on real data sets.
INFORMATION SCIENCES
(2021)
Article
Computer Science, Theory & Methods
Fernando Matia, Victor Jimenez, Biel P. Alvarado, Rodolfo Haber
Summary: This article revisits the fuzzy Kalman filter (FKF) and proposes a new method to represent uncertainty, avoiding implementation problems when propagating uncertainty and simplifying the computation steps of the Kalman filter. The FKF algorithm is reformulated with a new theoretical approach and validated tests are conducted in both simulation and a real mobile robot.
FUZZY SETS AND SYSTEMS
(2021)
Article
Chemistry, Analytical
Yarens J. Cruz, Marcelino Rivas, Ramon Quiza, Gerardo Beruvides, Rodolfo E. Haber
Review
Automation & Control Systems
Alberto Villalonga, Gerardo Beruvides, Fernando Castano, Rodolfo E. Haber
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2020)
Review
Automation & Control Systems
Alberto Villalonga, Elisa Negri, Giacomo Biscardo, Fernando Castano, Rodolfo E. Haber, Luca Fumagalli, Marco Macchi
Summary: This paper proposes a new framework for production scheduling optimization in cyber-physical production systems, utilizing the aggregation of multiple digital twins and combining local and global digital twins for decision-making supported by a fuzzy inference system. The framework is capable of detecting changes in the manufacturing process and making appropriate decisions for re-scheduling.
ANNUAL REVIEWS IN CONTROL
(2021)
Article
Chemistry, Analytical
Sebastian Bombinski, Joanna Kossakowska, Miroslaw Nejman, Rodolfo E. Haber, Fernando Castano, Robert Fularski
Summary: This paper outlines the specific needs and requirements of the aerospace industry in metal machining, introducing a concept of edge-computing-based production supervision system tailored for this industry. The system aims to provide real-time diagnostics and data sharing among technologists, line bosses, machine tool operators, and quality control personnel, addressing the unique demands of aerospace manufacturing.
Article
Computer Science, Interdisciplinary Applications
Yarens J. Cruz, Marcelino Rivas, Ramon Quiza, Alberto Villalonga, Rodolfo E. Haber, Gerardo Beruvides
Summary: This paper introduces an image classification approach based on an ensemble of convolutional neural networks and its application in a real industrial welding case study. Through an efficient search process, the method outperforms other strategies in detecting misalignment of metal sheets with maintained computational cost.
COMPUTERS IN INDUSTRY
(2021)
Article
Chemistry, Physical
Muhammad Umar Farooq, Saqib Anwar, M. Saravana Kumar, Abdullah AlFaify, Muhammad Asad Ali, Raman Kumar, Rodolfo Haber
Summary: This study investigates the influence of flushing attributes on wire electric discharge machining in the aeronautical industry. The optimized flushing mechanism was found to significantly improve dimensional precision and surface quality of the machined components, while reducing geometrical errors.
Article
Chemistry, Physical
Kiran Mughal, Mohammad Pervez Mughal, Muhammad Umar Farooq, Muhammad Qaiser Saleem, Rodolfo Haber Guerra
Summary: A novel solution using helical milling and nano fluid as a lubricant is investigated to improve the machining quality of CFRP/Ti6Al4V stacks in the aeronautical industry. The analysis of variance reveals that eccentricity, spindle speed (Ti), and tangential feed are the most significant parameters affecting diametric error for Ti6Al4V material, while tangential feed, spindle speed, and eccentricity are the most significant parameters for CFRP material. Furthermore, increasing the concentration of MoS2 can reduce the surface roughness of the Ti material, and spindle speed and lubricant have a potential influence on the processing temperature.
Article
Automation & Control Systems
Fernando Castano, Yarens J. Cruz, Alberto Villalonga, Rodolfo E. Haber
Summary: Nowadays, there is a growing emphasis on sustainability in industrial cyberphysical systems while maintaining productivity. This article proposes a data-driven method to predict the remaining useful life and classify faults using condition monitoring. By combining these outputs, a maintenance stop can be scheduled near the failure, improving sustainability without affecting productivity. A fuzzy decision-making strategy is developed to extend the useful life of electromechanical devices. A case study using bearing faults dataset confirms the effectiveness of the proposed methodology in achieving a trade-off between productivity and sustainability.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Materials Science, Multidisciplinary
Muhammad Umar Farooq, Saqib Anwar, Rizwan Ullah, Rodolfo Haber Guerra
Summary: In this study, the influence of process variables on stainless steel grade SS 316L manufactured through L-PBF is explored by high-speed turning. Parametric optimization is carried out to achieve desired response characteristics, such as lowering machining cost, minimizing carbon emissions, decreasing specific energy, enhancing tool life, and reducing surface roughness. This research can serve as a fundamental guideline for machinists in the metal processing industry and a basis for comparative analysis of sustainable machining routes.
JOURNAL OF MATERIALS RESEARCH AND TECHNOLOGY-JMR&T
(2023)
Article
Engineering, Electrical & Electronic
Wael M. Mohammed, Rodolfo E. Haber, Jose L. Martinez Lastra
Summary: This paper introduces the concept and implementation method of digital twin architecture, with a focus on the requirements of manufacturing processes. The approach is based on multi-layer and multi-level concepts, and uses multi-perspective view separation to provide a modular, scalable, reusable, interoperable, and composable approach for developing digital twins.
Proceedings Paper
Automation & Control Systems
A. Villalonga, E. Negri, L. Fumagalli, M. Macchi, F. Castano, R. Haber
Article
Computer Science, Information Systems
Daniel Rivas, Ramon Quiza, Marcelino Rivas, Rodolfo E. Haber
Proceedings Paper
Automation & Control Systems
Alberto Villalonga, Fernando Castano, Gerardo Beruvides, Rodolfo Laber, Stanislaw Strzelczak, Joanna Kossakowska
2019 23RD INTERNATIONAL CONFERENCE ON SYSTEM THEORY, CONTROL AND COMPUTING (ICSTCC)
(2019)
Article
Engineering, Civil
Fernando Castano, Rodolfo E. Haber, Wael M. Mohammed, Miroslaw Nejman, Alberto Villalonga, Jose L. Martinez Lastra
SMART STRUCTURES AND SYSTEMS
(2020)