Open access peer-reviewed chapter
Abstract
Multi-agent systems (MASs) are defined as a group of interacting entities or agents sharing a common environment that changes over time, with capabilities of perception and action, and the mechanisms for their coordination provide a modern perspective on systems that traditionally were regarded as centralized. The main characteristics of agents are learning and adaptation. In the last few years, MASs have received tremendous attention from scholars in different fields. However, there are still challenges faced by MASs and their integration with machine learning (ML) methods. The primary goal of the study is to provide a broad review of the current developments in the field of MASs combined with ML methods. First, we present features of MASs considering the ML perspective. Second, we provide a classification of applications of MASs combined with ML methods. Third, we present a density map of applications in E-learning, manufacturing, and commerce. We expect this study to serve as a comprehensive resource for researchers and practitioners in the area.
Keywords
- machine learning
- multi-agents
- simulation
- optimization
- neural networks
1. Introduction
At the beginning of the 1990s, agent-based programming became an important part of simulation [1]. Although there is currently no formal definition of what an agent is, the term is used to describe self-contained programs that can control their own actions based on perceptions of their operating environment [2]. The goal of agent-based programming is to create programs that intelligently interact with their environment. The software used to program agents has its origins in the field of artificial intelligence, especially in the subfield of distributed artificial intelligence [3, 4], whose objective is the study of the properties of agents and the design of interaction networks between them.
In the same 1990s, as is suggested in Ref. [5], computational agents are typically characterized by:
Autonomy: The agents had direct control of their actions and their internal state.
Social skills: Agents interacted with other agents through a computational language.
Reaction: Agents were able to perceive their environment and respond to it. The environment could be the physical world, a virtual world, or a simulated world that includes other agents.
Proactivity: Because agents reacted to their environment, they themselves had to take goal-oriented initiative.
A typical agent-based model contains the following four elements [6]:
Agents: Their attributes and environment.
Relationships between agents and methods of interaction.
A connectivity topology that defines how and with whom agents interact.
Agent environments: Agents live and interact with their environment and with other agents.
In most of the agent-based models, agents are able to move within their environment through sensors through which they perceive their local neighbors. Communication between agents is usually done by sending messages. For this action, the agents must be able to listen to the messages that come from their environment and send messages to the environment.
1.1 Multiagent systems
According to Garro et al. [7], a system comprising a number of possibly interacting agents is called a
1.2 Machine learning
Machine learning (ML) is a subset of artificial intelligence (AI) that concerns the development of algorithms, which allows the machine to learn via inductive inference based on observation data that represent incomplete information about statistical phenomena [9]. To carry out the learning process an algorithm is used based on examples of the task we want to solve (data) and letting the computer find patterns and make inferences that optimize the decision-making according to a user-defined objective [10]. Based on the training strategy, ML can be divided into three classical categories with different learning approaches: supervised learning, unsupervised learning, and reinforcement learning [10]. The first one includes classification and regression tasks, in the second one the widely used task is clustering, and the third one consists of the process of training a model on a series of actions that lead to a particular outcome, where the system receives rewards for performing well and punishment for performing poorly directly from its environment [10].
In the last years, MASs integrated with ML have received tremendous attention from scholars in different fields such as Computer Science, Engineering, Mathematics, Material Science, Neuroscience, Energy, Physics and Astronomy, Social Sciences, Environmental Sciences, Business, Management, and Accounting. The overview of MASs integrated with ML in these fields will be presented in the following sections. MASs have been used in areas of e-learning, manufacturing, and commerce combining mathematical methods, optimization methods, Markov processes, learning algorithms, and artificial intelligence techniques. The reinforcement learning method jointly with MASs has been applied in e-learning, manufacturing (multi-agent reinforcement learning), and commerce (machine learning and learning automation) areas. While deep reinforcement learning and adaptive learning have been applied jointly with MASs in manufacturing and commerce areas. However, there are still challenges faced by MASs and their integration with ML. The primary goal of the study is to provide a broad review of the current developments in the field of MASs combined with ML methods.
This chapter is prepared as follows: the methodology followed to carry out the state-of-the-art survey on various domains of multi-agent systems and machine learning is described in Section 2. Features of MASs considering the ML perspective are presented in Section 3. A density map of applications in e-learning, manufacturing, and commerce is provided in Section 4. Concluding remarks are drawn in Section 5.
2. Methodology
We followed the methodology proposed by Machi and McEvoy [11]. Figure 1 shows the steps for conducting the systematic literature review.
Figure 1.
The literature review model. Machi and McEvoy [
2.1 Step 1. Select a topic
The topic for this study was MASs combined with ML methods.
2.2 Step 2. Search the literature
The data for this study was extracted from the Scopus database (accessed on July 2022) based on the string
TITLE-ABS-KEY (
Figure 2.
Trends in the number of articles and chapter book by year that were published between 2017 and 2022.
Figure 3.
Trends in the number of articles and chapter book published annually by the top-five sources.
2.3 Step 3. Develop the argument
In this chapter, a broad review of the current developments in the field of MASs combined with ML methods is presented.
2.4 Step 4. Survey the literature
Figure 4 summarizes the top ten authors in terms of contribution to the number of papers and chapter book published on multi-agent systems and machine learning applications. Even though Yu [14, 15, 16, 17, 18] as coauthor is the main contributor in the area. He accounts for five articles, followed by Li [19, 20, 21, 22], Mohammed [23, 24, 25, 26], Wong [27, 28, 29, 30], Yap [27, 28, 29, 30], and Yaw [27, 28, 29, 30] with four articles each one. Figure 5 presents the coauthor collaboration network from articles and chapter book published during the 2017–2022 period time, on multi-agent systems and machine learning applications, based on full counting. Each circle represents an author. The size of a circle reflects the number of publications of the corresponding author. The distance between two circles indicates the relatedness of the authors [13]. Colors represent clusters of authors with strong coauthorship links. A total of 1,718 different authors are in the collaboration network.
Figure 4.
Documents by the author published between 2017 and 2022 on multi-agent systems and machine learning applications.
Figure 5.
Visualization by VOSviewerTM software of coauthor collaboration network from documents published between 2017 and 2022 on multi-agent systems and machine learning applications.
The top-ten fields in terms of applications on multi-agent systems and machine learning applications during the 2017–2022 period is showed in Figure 6. Even though Computer Science is the main field contributor, it only accounts for 37.4%, followed by Engineering (26.9%) and Mathematics (7.8%). The visualization in Figure 7, based on full counting, shows the co-occurrence network, distinct groups of keywords can be easily distinguished. Each circle represents a keyword. The size of a circle reflects the number of occurrences of the corresponding keyword. A total of 5,155 different words presented in the titles and abstracts of the 552 documents published between 2017 and 2022 were analyzed to establish the co-occurrence network, generating clusters associated with the research topic on multi-agent systems and machine learning applications. The colors used indicate the evolution of the time period of the different clusters. The blue cluster contained research topics published in 2019, while the green cluster contained research topics published in 2020, and the yellow cluster contained research topics published in 2021 and forward.
Figure 6.
Percentage of documents by subject fields that were published between 2017 and 2022.
Figure 7.
Visualization by VOSviewerTM software of word co-occurrence network built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications.
2.5 Step 5. Critique the literature
The critique of the scientific literature of MASs integrated with ML in Computer Science, Engineering, Mathematics, Material Science, Neuroscience, Energy, Physics and Astronomy, and Social Sciences fields will be presented in the following sections, highlighting their potentiality and limitations.
3. Review of state-of-the-art
3.1 Computer science
Computer Science turned out as the top field contributor as indicated by the retrieved data, with 440 documents on multi-agent systems and machine learning applications during the 2017–2022 time period. Figure 8 shows the terms with high co-occurrence frequencies in this field, distinguishing a color pallet from blue to yellow based on the publication year. In Figure 8, each term is represented by a circle, where the diameter of the circle and size of its label represents the frequency of the term, and its proximity to another term indicates the degree of relatedness of the two terms.
Figure 8.
Visualization by VOSviewerTM software of word co-occurrence network built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications in the Computer Science field.
3.1.1 Potentiality
The network in Figure 8 clearly indicates that multi-agent systems and machine learning have received tremendous attention from scholars. In 2019, the research was focused on topics such as embedded systems, robots, and forecasting, whereas in 2020, it was focused on topics such as intelligent agents, decision-making systems, optimization, IoT, stochastics systems scheduling, wireless sensor networks, compute games, and energy and efficiency of systems, which were broadly studied. More recently, in 2021 resource allocation, heuristics algorithms, decision trees, real-time systems, distributed optimization, and predictive analysis jointly with multi-agent systems were applied to study street traffic, manufacture, and cognitive radio.
3.1.2 Limitations
ML algorithms and multi-agent systems showed potentially significant combination/integration in the Computer Sciences field. However, the studies in this field had several limitations. First, they had a small number of applications. Second, the ML algorithms are limited to deep reinforcement. Finally, it is unclear which software is used to integrate multi-agent systems and ML algorithms.
3.2 Engineering
Engineering turned out as the second contributor as indicated by the retrieved data, with 317 documents on MASs and ML applications between 2017 and 2022. Figure 9 shows the terms with high co-occurrence frequencies in this field. In Figure 9, each term is represented by a circle, where the diameter of the circle and size of its label represent the frequency of the term, and its proximity to another term indicates the degree of relatedness of the two terms.
Figure 9.
Visualization by VOSviewerTM software of word co-occurrence network built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications in the Engineering field.
3.2.1 Potentiality
As is depicted in Figure 9, MASs researchers are very enthusiastic about learning algorithms to study vehicular networks, traffic control, 5G mobile communications networks, image communication systems, multi-robots systems, wireless sensor networks, manufacture, digital storage, electric power utilization, cyber-physical systems, embedded systems, electric vehicles, electric power transmission, and e-learning, and have proposed reinforcement learning and deep reinforcement learning to be integrated to MASs.
3.2.2 Limitations
Although, MASs interacting with reinforcement learning and deep reinforcement learning algorithms showed potentially significant application in the Engineering field. However, there were several potential limitations, such as small minority of optimization techniques, traditional simulation approaches, and little information about the integration of optimization algorithms and simulation software.
3.3 Mathematics
Mathematics field turned out as the third contributor as indicated by the retrieved data, with 92 documents on MASs and ML applications between 2017 and 2022.
3.3.1 Potentiality
The number of publications increased noticeably in the past two years. The scientific landscape of main research areas of MASs and ML applications in the Mathematics field are bibliometrically explored by way of co-occurrence term map presented in Figure 10. In Figure 10, each term is represented by a circle, where the diameter of the circle and size of its label represent the frequency of the term, and its proximity to another term indicates the degree of relatedness of the two terms.
Figure 10.
Visualization by VOSviewerTM software of word co-occurrence network built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications in the Mathematics field.
The most used theoretical tools were deep learning, neural networks, intelligent agents, Markov process, scheduling algorithms, q-learning algorithms, decision trees, and game theory. While the most important application areas were decision making, job scheduling, power control, cloud computing, energy efficiency, fertilizers, e-commerce, speech processing, cooperative behaviors, quality of service, resource allocation, forecasting, and manufacturing.
3.3.2 Limitations
A considerable amount of literature has been published on multi-agent systems and machine learning applications in the Mathematics field. However, the reinforcement learning method has been recently used in specific areas such as manufacturing and job scheduling supporting decision making. Optimization algorithms based on warm intelligence were used in the past as well as supervised learning methods. Much of the recent literature in this field has limited applications that are centered on cloud computing.
3.4 Materials science
Materials Science field turned out as the fourth contributor as indicated by the retrieved data, with 63 documents on MASs and ML applications between 2017 and 2022. Figure 11 shows the mapping and clustering of terms over time. A co-occurrence network was built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications in the Materials Science field. In Figure 11, each term is represented by a circle, where the diameter of the circle and size of its label represent the frequency of the term, and its proximity to another term indicates the degree of relatedness of the two terms. The network can be seen to contain four clusters of co-occurring terms. The royal blue cluster is predominately associated with documents published in 2019, the turquoise cluster is associated with documents published in 2020, the yellow-green cluster is associated with documents, and the yellow cluster is associated with documents recently published in 2022.
Figure 11.
Visualization by VOSviewerTM software of word co-occurrence network built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications in the Materials Science field.
3.4.1 Potentiality
For the period of 2017–2022, multi-agent reinforcement learning, deep learning, deep reinforcement learning, complex networks, multi-agent systems, stochastic systems, and game theory remained among the top major topics. It has been demonstrated that deep reinforcement learning methods have been combined with game theory and deep neural network techniques. Additionally, complex networks have considered computational complexity.
3.4.2 Limitations
Although there are many studies, the research in multi-agent systems and machine learning applications in the Materials Science field remains limited. First, the application areas have been centered on traffic studies, scheduling, Internet of Things, crowd simulation, and more recently in energy efficiency and wireless networks. Second, the learning algorithms are mainly based on deep learning. Additionally, optimization is based on the particle swarm algorithm, mainly.
3.5 Neuroscience
Neuroscience field turned out as the fifth contributor as indicated by the retrieved data, with 33 documents on MASs and ML applications between 2017 and 2022. Figure 12 depicts the terms with high co-occurrence frequencies in this field. In Figure 12, each term is represented by a circle, where the diameter of the circle and size of its label represent the frequency of the term, and its proximity to another term indicates the degree of relatedness of the two terms.
Figure 12.
Visualization by VOSviewerTM software of word co-occurrence network built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications in the Neuroscience field.
3.5.1 Potentiality
The number of publications increased noticeably in the past two years on multi-agent systems and machine learning applications in the Neuroscience field. The terms on the left of the network shown in Figure 12 represent the contributions developed in 2018. Here, mathematical models, simulation models, as well as analytic and optimization methods were broadly used. The group of terms at the center of the diagram characterizes the contributions developed between 2019 and 2020. Here, artificial neural networks, nonlinear systems, graph theory, and simulation were the theoretical tool mainly used. The right of the diagram contains terms related to contributions in the last year. Here, the learning and reinforcement algorithms have been broadly applied.
3.5.2 Limitations
ML algorithms and multi-agent systems showed potentially significant combination/integration in the Neurosciences field. However, the studies in this field had several limitations. First, the applications are centered on autonomous systems that learn using deep learning and reinforcement learning algorithms. Second, the reward concept is minimum exploited in reinforcement learning. The software to implement multi-agent systems has little attention. Finally, computer simulation was relevant in studies developed in 2018 but is not included in the studies recently developed.
3.6 Energy
Energy field turned out as the sixth contributor as indicated by the retrieved data, with 32 documents on MASs and ML applications between 2017 and 2022. Figure 13 shows the terms with high co-occurrence frequencies in this field. In Figure 13, each term is represented by a circle, where the diameter of the circle and size of its label represent the frequency of the term, and its proximity to another term indicates the degree of relatedness of the two terms.
Figure 13.
Visualization by VOSviewerTM software of word co-occurrence network built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications in the energy field.
3.6.1 Potentiality
A number of specific research topics show significant active growth and may be considered to be emerging topics on multi-agent systems and machine learning applications in the Energy field. Reinforcement learning is the dominant algorithm used followed by deep learning and deep reinforcement learning. Optimization is based on learning algorithms. The main applications in this field are electric machine control, electric vehicles, housing, fertilizers, smart grid, electric power transmission networks, microgrids, energy management, and energy utilization.
3.6.2 Limitations
There are two major limitations in the applications of multi-agent systems and machine learning in the Energy field. First, the more recent applications are centered on model predictive control. Second, the electric cost is evaluated in a few studies.
3.7 Physics and astronomy
Physics and Astronomy field turned out as the seventh contributor as indicated by the retrieved data, with 31 documents on MASs and ML applications between 2017 and 2022. Figure 14 illustrates the network of terms with high co-occurrence frequencies in this field. In Figure 14, each term is represented by a circle, where the diameter of the circle and size of its label represent the frequency of the term, and the distance between any two possible terms reflects the relatedness of the terms as closely as possible. In general, the stronger the relationship between two terms, the smaller the distance between the terms on the map. The network can be seen to contain four clusters of co-occurring terms.
Figure 14.
Visualization by VOSviewerTM software of word co-occurrence network built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications in the Physics and Astronomy field.
3.7.1 Potentiality
After careful analysis, in 2019, three topics can be distinguished (the royal blue biggest circles in Figure 14). The first is learning algorithms, the second topic is sensor nodes, and the third topic is game theory.
In 2020, six topics (the turquoise biggest circles in Figure 14) were the most often discussed of the last three years. The observation that reinforcement learning is one of the most occurring terms in the Physics and Astronomy field does not come as a surprise Internet of Things.
In 2021, six topics (the yellow biggest circles in Figure 14) were the most predominant of the last three years. The prominence of neural networks may be explained by the fact that is linked directly to machine learning approaches, constituent models, and multilayer neural networks.
3.7.2 Limitations
Although, MASs interacting with reinforcement learning and deep reinforcement learning algorithms showed potentially significant application in Physics and Astronomy field. However, there were several potential limitations, such as small minority of simulation and optimization models, learning algorithms, and software.
4. Applications of MASs combined with ML
Density map assists in the understanding of active growth areas, research trends, emerging topics, and hot topics in MASs combined with ML. Figures 15–17 illustrate the density map of applications in E-learning, manufacturing, and commerce, respectively.
Figure 15.
Density map by VOSviewerTM software of word co-occurrence network built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications in E-learning.
Figure 16.
Density map by VOSviewerTM software of word co-occurrence network built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications in manufacturing.
Figure 17.
Density map by VOSviewerTM software of word co-occurrence network built using words present in titles and abstracts of documents published between 2017 and 2022 on multi-agent systems and machine learning applications in commerce.
5. Conclusions
The primary goal of the study was to provide a broad review of the current developments in the field of MASs combined with ML methods. The trend on MASs and ML is the use of reinforcement learning algorithms integrated with optimization and simulation models. Artificial neural networks, nonlinear systems, and graph theory are the theoretical tool mostly used. We want to emphasize that it is unclear which software is used to integrate multi-agent systems and ML algorithms. In most studies, optimization algorithms were based on warm intelligence.
Acknowledgments
The authors appreciate the partial support by UNAM- PAPIME PE 104022. The authors also gratefully acknowledge all the time and support received from IntechOpen.
References
- 1.
Maes P. Agents that reduce work and information overload. Communications of the ACM. 1994; 37 :31-40 - 2.
Huhns M, Singh MP. Readings in Agents. 1st ed. San Mateo, CA: Morgan Kaufmann; 1998 - 3.
Bond AH, Gasser L. Readings in Distributed Artificial Intelligence. Los Altos, CA, USA: Morgan Kaufmann; 1998 - 4.
Chaib-draa B, Moulin B, Mandiau R, Millot P. Trends in distributed artificial intelligence. Artificial Intelligence Review. 1992; 6 :35-66 - 5.
Wooldridge M, Jennings NR. Intelligent agents: Theory and practice. The Knowledge Engineering Review. 1995; 10 :115-152 - 6.
Macal C, North M. Introductory tutorial: Agent-based modeling and simulation. In: Proceedings of the 2011 Winter Simulation Conference. Phoenix, Arizona, USA: IEEE; 2011. pp. 1468-1468 - 7.
Garro A, Mühlhäuser M, Tundis A, Baldoni M, Baroglio C, Bergenti F, et al. Intelligent agents: Multi-agent systems. Encyclopedia of Bioinformatics and Computational Biology. 2019; 1 :315-320 - 8.
Al-Jumaily A, Al-Jaafreh M. Multi-agent system concepts, Theory and Applications phases. In: Buchli J, editor. Mobile Robots, Moving Intelligence. London: InTech; 2006. pp. 369-392 - 9.
Cen L, Dong M, Li Zhu H, Yu L, Chan P. Machine learning methods in the application speech emotion recognition. In: Zhang Y, editor. Application of Machine Learning. London: InTech; 2010 - 10.
Haidine A, Zahra Salmam F, Aqqal A, Dahbi A. Artificial intelligence and machine learning in 5G and beyond: A survey and perspectives. In: Haidine A, editor. Moving Broadband Mobile Communications Forward. Intelligent Technologies for 5G. London: InTech; 2021 - 11.
Machi LA, Mcevoy BT. The Literature Review. California: Corwin Press; 2009 - 12.
Van Eck NJ, Waltman L. Software survey: VOSviewer, a computer program for bibliometric mapping. Scientometrics. 2010, 2010; 84 :523-538 - 13.
Van Eck NJ, Waltman L. Visualizing bibliometric networks. In: Ding Y, Rousseau R, Wolfram D, editors. Measuring Scholarly Impact: Methods and Practice. Belin: Springer; 2014. pp. 285-320 - 14.
Li J, Yang B, Yu T. Distributed deep reinforcement learning-based coordination performance optimization method for proton exchange membrane fuel cell system. Sustainable Energy Technologies and Assessments. 2022; 50 :101284 - 15.
Li J, Yu T. Large-scale multi-agent deep reinforcement learning-based coordination strategy for energy optimization and control of proton exchange membrane fuel cell. Sustainable Energy Technologies and Assessments. 2021; 48 :101752 - 16.
Chen Y, Zhang X, Guo L, Yu T. Optimal carbon-energy combined flow in power system based on multi-agent transfer reinforcement learning. Gaodianya Jishu/High Voltage Engineering. 2019; 45 :3 - 17.
Cheng L, Yu T. Exploration and exploitation of new knowledge emergence to improve the collective intelligent decision-making level of web-of-cells with cyber-physical-social systems based on complex network modeling. IEEE Access. 2018; 6 - 18.
Zhang X, Chen Y, Yu T, Yang B, Qu K, Mao S. Equilibrium-inspired multiagent optimizer with extreme transfer learning for decentralized optimal carbon-energy combined-flow of large-scale power systems. Applied Energy. 2018; 189 :157-176 - 19.
Liu C, Shi Y, Li H, Du W. Accelerated dual averaging methods for decentralized constrained optimization. IEEE Transactions on Automatic Control. 2022. DOI: 10.1109/TAC.2022.3173062 - 20.
Sun B, Hu J, Xia D, Li H. A distributed stochastic optimization algorithm with gradient-tracking and distributed heavy-ball acceleration. Frontiers of Information Technology and Electronic Engineering. 2021; 22 :11 - 21.
Li H, Hu J, Ran L, Wang Z, Lü Q , Du Z, et al. Decentralized dual proximal gradient algorithms for non-smooth constrained composite optimization problems. IEEE Transactions on Parallel and Distributed Systems. 2021; 32 :103 - 22.
Hu J, Ran L, Du Z, Li H. Decentralized stochastic optimization algorithms using uncoordinated step-sizes over unbalanced directed networks. Signal Processing. 2021; 32 :10 - 23.
Mohammed MA, Elhoseny M, Abdulkareem KH, Mostafa SA, Maashi MS. A multi-agent feature selection and hybrid classification model for parkinson’s disease diagnosis. ACM Transactions on Multimedia Computing, Communications and Applications. 2021; 17 :2 - 24.
Mohammed MA, Ibrahim DA, Salman AO. Adaptive intelligent learning approach based on visual anti-spam email model for multi-natural language. Journal of Intelligent Systems. 2021; 30 :1 - 25.
Elhoseny M, Mohammed MA, Mostafa SA, Abdulkareem KH, Maashi MS, Garcia-Zapirain B, et al. A new multi-agent feature wrapper machine learning approach for heart disease diagnosis. Computers, Materials and Continua. 2021; 67 :1 - 26.
Mostafa SA, Mustapha A, Mohammed MA, Hamed RI, Arunkumar N, Abd Ghani MK, et al. Examining multiple feature evaluation and classification methods for improving the diagnosis of Parkinson’s disease. Cognitive Systems Research. 2019; 54 :90-99 - 27.
Yaw CT, Yap KS, Wong SY, Yap HJ, Paw JKS. Enhancement of neural network based multi agent system for classification and regression in energy system. IEEE Access. 2020; 8 :163026-163043 - 28.
Yaw CT, Wong SY, Yap KS, Tan CH. Extreme learning machine neural networks for multi-agent system in power generation. International Journal of Engineering and Technology (UAE). 2018; 7 :4 - 29.
Yaw CT, Wong SY, Yap KS, Yap HJ, Amirulddin UAU, Tan SC. An ELM based multi-agent system and its applications to power generation. Intelligent Decision Technologies. 2018; 12 :2 - 30.
Yaw CT, Wong SY, Yap KS, Yap HJ, Amirulddin UAU, Tan SC. An ELM based multi-agent system and its applications to power generation. Intelligent Decision Technologies. 2017; 11 :3