Article
Computer Science, Artificial Intelligence
Jie Cao, Jianlin Zhang, Fuqing Zhao, Zuohan Chen
Summary: A novel algorithm named MOEA/D-TS is proposed in this paper, which effectively solves multi-objective optimization problems through two-stage evolution strategies. The performance of the algorithm is validated in real world problems and shows advantages in terms of convergence and diversity over other algorithms.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Artificial Intelligence
Ying Xu, Huan Zhang, Lei Huang, Rong Qu, Yusuke Nojima
Summary: This research investigates the grid-based decomposition methods in multi-objective optimization to address the issues of diversity and convergence. A new concept of Pareto Front grid and a statistical analysis-based nadir point estimation strategy are proposed to improve computational efficiency. Furthermore, a novel grid-based knee point selection method is proposed. Experimental analysis demonstrates the effectiveness of the proposed PFG-MOEA algorithm.
APPLIED SOFT COMPUTING
(2023)
Article
Automation & Control Systems
Shuwei Zhu, Lihong Xu, Erik D. Goodman, Zhichao Lu
Summary: The article proposes a new multiobjective evolutionary algorithm based on the generalization of Pareto optimality, which uses the (M-1)-GPD framework to promote both convergence and diversity. Research shows that this algorithm is competitive on various benchmark problems and outperforms other methods on three real-world problems.
IEEE TRANSACTIONS ON CYBERNETICS
(2022)
Article
Computer Science, Information Systems
Jinlong Zhou, Juan Zou, Shengxiang Yang, Jinhua Zheng, Dunwei Gong, Tingrui Pei
Summary: This paper proposes niche-based and angle-based selection strategies for many objective evolutionary optimization, which have been shown to be competitive and scalable to handle constrained many-objective optimization problems in experimental studies.
INFORMATION SCIENCES
(2021)
Article
Computer Science, Artificial Intelligence
Derya Deliktas, Ender Ozcan, Ozden Ustun, Orhan Torkul
Summary: The study introduces evolutionary algorithms to solve the bi-objective flexible job shop scheduling problem and compares their performance across various configurations. The transgenerational memetic algorithm using weighted sum method outperforms others and achieves the best-known results for almost all instances of bi-objective flexible job shop cell scheduling.
APPLIED SOFT COMPUTING
(2021)
Article
Computer Science, Artificial Intelligence
Qian Wang, Qinghua Gu, Lu Chen, Yueping Guo, Naixue Xiong
Summary: This paper proposes an improved multi-objective evolutionary algorithm with global and local cooperative mechanisms for complicated bi-objective optimization problems. The overall optimization is carried out through the coordination of global and local search phases. The proposed algorithm shows great potential for addressing complex bi-objective optimization problems.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Information Systems
Mingjing Wang, Xiaoping Li, Yong Dai, Long Chen, Huiling Chen, Ruben Ruiz
Summary: Researchers have developed a method called Copula Incremental Learning (CIL) to improve the performance of the MOEA/D algorithm in problems with irregular Pareto Fronts (PFs) by generating non-uniform direction vectors. They also employ the Niche Hierarchical Selection (NHS) method to construct the neighborhood structure and prevent duplicate solutions. The use of convergence-guided direction (CGD) ensures efficiency by approximating irregular PFs. Statistical analysis shows that this method outperforms other competitive algorithms, particularly in handling multi-objective optimization problems with irregular PFs.
INFORMATION SCIENCES
(2023)
Article
Computer Science, Artificial Intelligence
Juan Zou, Zhenghui Zhang, Jinhua Zheng, Shengxiang Yang
Summary: This paper proposes a co-guided evolutionary algorithm that combines the merits of dominance and decomposition to balance convergence and diversity in the evolutionary process. Experimental results demonstrate the superiority and versatility of this method in MaOPs.
KNOWLEDGE-BASED SYSTEMS
(2021)
Article
Computer Science, Artificial Intelligence
Li Li, Guangpeng Li, Liang Chang
Summary: Evolutionary multi-objective optimization problems have gained attention in the evolutionary computing community. Existing methods for improving the scalability of multi-objective evolutionary algorithms (MOEAs) have difficulty balancing convergence and diversity as the number of objectives increases. This paper proposes a self-adaptive stochastic ranking method (SSR) and an improved density estimation strategy (ISDE) to address this issue. Experimental results show that the proposed algorithm performs competitively compared to state-of-the-art MOEAs.
Article
Automation & Control Systems
Qinqin Fan, Yilian Zhang, Ning Li
Summary: The paper introduces an automatic selection strategy of multiobjective evolutionary algorithms based on performance indicators (MOEAS-PI). This strategy can effectively improve the efficiency and robustness of solving multiobjective optimization problems.
IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Maroua Grid, Leyla Belaiche, Laid Kahloul, Saber Benharzallah
Summary: Multi-Objective Optimization Evolutionary Algorithms (MOEAs) are proposed heuristic methods for solving Multi-objective Optimization Problems (MOPs). While Dynamic Multi-Objective Evolutionary Algorithm (DMOEA) has dynamic population size, the Parallel Dynamic Multi-Objective Evolutionary Algorithm (PDMOEA) aims to improve efficiency in terms of objective space, computational time, and convergence to desired population size compared to DMOEA. Additionally, a new formula is established to determine the suitable number of processes required in PDMOEA for optimal solutions.
INTERNATIONAL ARAB JOURNAL OF INFORMATION TECHNOLOGY
(2022)
Article
Multidisciplinary Sciences
Aamir Rizwan, Iftikhar Ahmad, Muhammad Asif Zahoor Raja, Muhammad Shoaib
Summary: A novel stochastic numerical computing method is introduced for nonlinear thin film flow (TFF) system in computational fluid dynamics, utilizing polynomial splines and evolutionary computing. The proposed approach, CSA-GA-SQP, is shown to be efficient, reliable and stable alternative for solving stiff nonlinear systems in complex scenarios of TFF models.
ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING
(2021)
Article
Engineering, Chemical
Hongbo Jiao, Huaibin Wei, Qi Yang, Min Li
Summary: The CFD-MOEA/D algorithm is proposed for flood control operation of large-scale reservoirs, showing better performance than traditional algorithms. By using a decomposition method, it obtains non-dominated solutions with higher water levels and outperforms the NSGA-II algorithm. The optimal dispatching scheme of the algorithm matches the actual reservoir dispatching, improving scheduling efficiency.
Article
Computer Science, Artificial Intelligence
Suraj S. Meghwani, Manoj Thakur
Summary: This study proposes an adaptive strategy to modify scalarizing weights in multi-objective evolutionary algorithm based on Decomposition (MOEA/D) after regular intervals by assessing the crowdedness of solutions using crowding distance operator. The experiments show that this strategy improves the convergence and diversity of solutions on approximated Pareto-Front, and the proposed algorithm performs better than other state-of-the-art multi-objective algorithms on most benchmark problems.
APPLIED INTELLIGENCE
(2021)
Article
Computer Science, Artificial Intelligence
Josiel Neumann Kuk, Richard Aderbal Goncalves, Lucas Marcondes Pavelski, Sandra Mara Guse Scos Venske, Carolina Paula de Almeida, Aurora Trinidad Ramirez Pozo
Summary: This paper analyzes different multi-objective evolutionary algorithms for handling the Environmental/Economic Load Dispatch problem and evaluates the impact of a repair procedure on algorithm performance. Experiments show that NSGA-II with repair procedure demonstrates consistency and outperforms other approaches in most instances.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Operations Research & Management Science
Bhupinder Singh Saini, Michael Emmerich, Atanu Mazumdar, Bekir Afsar, Babooshka Shavazipour, Kaisa Miettinen
Summary: This paper introduces a novel concept to address multiobjective optimization problems with computationally expensive function evaluations. The proposed interactive method, O-NAUTILUS, combines trade-off free search and navigation, utilizing uncertainty quantification from surrogate models to approximate an optimistic Pareto optimal set.
JOURNAL OF GLOBAL OPTIMIZATION
(2022)
Article
Operations Research & Management Science
Javad Koushki, Kaisa Miettinen, Majid Soleimani-damaneh
Summary: In this paper, an interactive algorithm called LR-NIMBUS is developed to assist decision makers in finding a preferred lightly robust efficient solution for uncertain multiobjective optimization problems. The algorithm extends the interactive NIMBUS method and incorporates decision maker preferences through objective function classification. By solving an augmented weighted achievement scalarizing function, a lightly robust efficient solution is generated with tractability demonstrated for important classes of objective functions and uncertainty sets. As an illustrative example, a robust optimization problem in stock investment (portfolio selection) is modeled and solved.
JOURNAL OF GLOBAL OPTIMIZATION
(2022)
Article
Computer Science, Artificial Intelligence
Bekir Afsar, Ana B. Ruiz, Kaisa Miettinen
Summary: This paper highlights the importance of solving multiobjective optimization problems with interactive methods and comparing different methods to find the most suitable one. It introduces a new artificial decision maker (ADM-II) that can handle different types of preference information and assess the performance of interactive evolutionary methods. By considering the learning and decision phases separately, ADM-II can generate preference information in different ways to reflect the nature of each phase.
COMPLEX & INTELLIGENT SYSTEMS
(2023)
Editorial Material
Operations Research & Management Science
Karl-Heinz Kuefer, Kaisa Miettinen, Stefan Ruzika, Serpil Sayin
Article
Automation & Control Systems
Giovanni Misitano, Bekir Afsar, Giomara Larraga, Kaisa Miettinen
Summary: Interactive multiobjective optimization methods incorporate decision maker preferences and provide explanations for solution process, with R-XIMO method supporting decision makers to understand trade-offs in the problem and improve desired objectives by suggesting impairing another objective.
AUTONOMOUS AGENTS AND MULTI-AGENT SYSTEMS
(2022)
Article
Forestry
Babooshka Shavazipour, Dmitry Podkopaev, Kaisa Miettinen
Summary: Sustainable environmental management often involves long-term time horizons and multiple conflicting objectives and is affected by different sources of uncertainty. This paper proposes a novel interactive multi-scenario multiobjective approach to support decision-making and trade-off analysis in sustainable forest landscape planning under multiple sources of uncertainty.
CANADIAN JOURNAL OF FOREST RESEARCH
(2022)
Article
Computer Science, Interdisciplinary Applications
Adhe Kania, Juha Sipila, Giovanni Misitano, Kaisa Miettinen, Jussi Lehtimaki
Summary: This study addresses the challenges of unpredictable demand and proposes a multiobjective optimization model to integrate a lot sizing problem with safety strategy placement. The proposed model considers four objective functions and is solved using the E-NAUTILUS method. The results demonstrate that the model can help decision makers find the best balance among conflicting objectives.
COMPUTERS & INDUSTRIAL ENGINEERING
(2022)
Article
Computer Science, Artificial Intelligence
Atanu Mazumdar, Tinkle Chugh, Jussi Hakanen, Kaisa Miettinen
Summary: This article proposes probabilistic selection approaches that utilize the uncertainty information of the Kriging models to improve the solution process in offline data-driven multiobjective optimization. The experimental results show that these approaches can produce solutions with a greater hypervolume and a lower root mean squared error compared to generic approaches and a transfer learning approach that do not use uncertainty information.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION
(2022)
Article
Management
Bekir Afsar, Johanna Silvennoinen, Giovanni Misitano, Francisco Ruiz, Ana B. Ruiz, Kaisa Miettinen
Summary: Interactive multiobjective optimization methods operate iteratively, allowing decision makers to provide preference information and generate desired solutions. Different methods vary in technical aspects and preference information used, making it challenging to select the most suitable method. Published research lacks specific information on conducted experiments, impeding replication. We propose a novel questionnaire and experimental design for comparing methods and develop user interfaces for a sustainability problem with multiple objectives.
JOURNAL OF THE OPERATIONAL RESEARCH SOCIETY
(2023)
Article
Automation & Control Systems
Bhupinder Singh Saini, Debalay Chakrabarti, Nirupam Chakraborti, Babooshka Shavazipour, Kaisa Miettinen
Summary: This paper tackles the challenges of solving real-life data-driven multiobjective optimization problems, involving preprocessing, modelling, formulation, and decision support. It focuses on optimizing the composition of microalloyed steels to achieve desired mechanical properties. The proposed MultiDM/IOPIS algorithm combines multiobjective evolutionary algorithms and scalarization functions to enable meaningful decision-making for multiple objective functions and decision makers. Through the use of the DESDEO framework, the methodology successfully provides microalloyed steel compositions that satisfy both decision makers.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Adhe Kania, Bekir Afsar, Kaisa Miettinen, Juha Sipila
Summary: We propose DESMILS, a decision support approach that tackles multi-item lot sizing problems with a large number of items using single-item multiobjective lot sizing models. DESMILS considers multiple conflicting objective functions and incorporates decision maker preferences to find the most preferred Pareto optimal solutions. Through clustering, DESMILS treats items in a cluster utilizing preferences provided for a representative item. This approach reduces the decision maker's workload and time while still achieving acceptable solutions.
JOURNAL OF INTELLIGENT MANUFACTURING
(2023)
Article
Cardiac & Cardiovascular Systems
Arto J. Hautala, Babooshka Shavazipour, Bekir Afsar, Mikko P. Tulppo, Kaisa Miettinen
Summary: This study evaluated the applicability of machine learning tools for predicting healthcare costs in patients with acute coronary syndrome based on known risk markers, and found that depression score is the most significant predictor of healthcare costs.
CARDIOVASCULAR DIGITAL HEALTH JOURNAL
(2023)
Article
Engineering, Multidisciplinary
Jana Burkotova, Pouya Aghaei Pour, Tomas Kratky, Kaisa Miettinen
Summary: This article introduces a surrogate-assisted evolutionary interactive multiobjective optimization method applied to pump stator design. The preferences of a decision maker are iteratively incorporated into the solution process, demonstrating the advantages of the interactive method in reducing computation time and finding preferred solutions. The decision maker expressed satisfaction with the interactive solution process, and the final solution accurately reflected his preferences. Importantly, this method could save days of computation time.
ENGINEERING OPTIMIZATION
(2023)
Proceedings Paper
Computer Science, Artificial Intelligence
Giomara Larraga, Kaisa Miettinen
Summary: Multiobjective optimization aims to help decision-makers find satisfying solutions for problems with multiple conflicting objectives. This article proposes an interactive version of MOEA/D that incorporates three types of preference information to provide more flexibility in guiding the search. Applied to a river pollution problem, the method demonstrates its potential in supporting decision-makers to find satisfying solutions.
PROCEEDINGS OF THE 2022 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE COMPANION, GECCO 2022
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Atanu Mazumdar, Stefan Otayagich, Kaisa Miettinen
Summary: This paper proposes a fully modular physical user interface for inputting preference information in solving multiobjective optimization problems. The interface can be used with any computer and employs web-based visualizations. The potential of the physical interface is demonstrated by solving a real-world problem using an interactive decomposition-based multiobjective evolutionary algorithm.
PROCEEDINGS OF THE 2022 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE COMPANION, GECCO 2022
(2022)
