Search

Computational Intelligence (CI) is a more efficient paradigm for solving real-world problems in uncertain conditions. The traditional CI approaches are not capable to provide the complete and sufficient solutions for problems. Therefore, new techniques are necessary to efficiently solve these issues seriously. New techniques, such as Emergent Intelligence (EI), Multi-Agent System (MAS), etc., provide robust, generic, flexible, and self-organised to solve complex real-world problems. In this paper, we discuss Emergent Intelligence (EI) and its uniqueness in solving problems in an uncertain environment. We also discuss EI, Swarm Intelligence (SI) and MultiAgent System (MAS)-based problem-solving in an uncertain environment and compared their performance. We have considered two different problems: job shop scheduling using EI and MAS and route establishment for routing using MAS, SI and EI in an uncertain environment. Each problem is categorically analysed and solved step by step using MAS, SI and EI in a dynamic environment. We measure the performance of these three methods by varying the number of agents, tasks and time. Performance measures are compared and shown to demonstrate the importance of EI over MAS and SI for solving problems in an uncertain environment. [ABSTRACT FROM AUTHOR]

The current century has brought an unimaginable growth in information and communications technology (ICT) and needs of enormous computing. The advancements in computer hardware and software particularly helped fuel the requirements of human beings, and revolutionized the smart products as an outcome. The advent of wearable devices from their development till successful materialisation has only taken less than a quarter of a century. The huge benefits of these smart wearable technologies cannot be fully enjoyed until and unless the reliability of a complete system is ensured. The reliability can be increased by the consistent advancements in hardware and software in parallel. User expectations actually are the challenges that keep the advancements alive while improving at an unmatchable pace. The future of wearable and other smart devices depends on whether they can provide a timely solution that is reliable, richer in resources, smaller in size, and cheaper in price. This paper addresses the threats and opportunities in the development and the acceptance of immersive and wearable technologies. The hardware and software challenges for the purpose of development are discussed to demonstrate the bottlenecks of the current technologies and the limitations that impose those bottlenecks. For the purpose of adoption, social and commercial challenges related to innovation and acceptability are discussed. The paper proposes guidelines that are expected to be applicable in several considerable applications of wearable technologies, for example, social networks, healthcare, and banking. [ABSTRACT FROM PUBLISHER]

The objective of this work was the proposal, design, construction and validation of a mobile health system for dietetic monitoring and assessment, called SapoFit. This application may be personalized to keep a daily personal health record of an individual's food intake and daily exercise and to share this with a social network. The initiative is a partnership with SAPO - Portugal Telecom. SapoFit uses Web services architecture, a relatively new model for distributed computing and application integration. SapoFit runs on a range of mobile platforms, and it has been implemented successfully in a range of mobile devices and has been evaluated by over 100 users. Most users strongly agree that SapoFit has an attractive design, the environment is user-friendly and intuitive, and the navigation options are clear. [ABSTRACT FROM AUTHOR]

Multiple Criteria Decision Making (MCDM) methods have been used in various planning applications to classify, select, or order alternatives aggregated within a decision-making process. However, methods such as the analytic hierarchy process (AHP) in its traditional form, linearized AHP, and fuzzy-AHP extent analysis require the generation of paired comparison matrices associated with the alternatives for decision-making purposes. This work proposes a hybrid decision-making model based on the combination of the judgment matrix of the AHP method and an expert fuzzy system. A real case study of a concessionaire in Brazil is analyzed to validate the proposed model. The case study has 4 feeders that must be built. The traditional AHP method was applied, with the evaluation of 4 experts who produced a criteria judgment matrix containing the 10 analyzed criteria. After performing 240 comparisons between alternatives. The traditional AHP indicated priority indices of 50.35%, 23.98%, 14.25%, and 11.42%, respectively, for Feeders 1, 2, 3 and 4. The proposed hybrid AHP-Fuzzy method, using the same criteria judgment matrix elaborated by the experts, indicated priority indices of 40.36%, 25.04%, 24.82%, and 9.78%, respectively, for Feeders 1, 2, 3 and 4. The order of the priority obtained by the two methods was the same with the advantage that the proposed approach did not need to generate paired comparison matrices, contributing to minimize the dependence on human intervention and speed up the decision-making process. In the case study, the methods were also applied to linearized AHP and fuzzy-AHP extent analysis, which provided priority indices similar to those of the other methods, keeping the order of prioritization. [ABSTRACT FROM AUTHOR]

Clustering of electrical load curves is an important resource used in the analysis of Electric Power Systems, as it extracts characteristics from the behavior of the load. Real data of load curve presents outliers and acquisition problems that are not suitable for clustering. This research proposes a hybrid algorithm for load curve filtering and clustering. An iterative process using the Hampel filter and the Wavelet Transform was applied to filter the data from the load curve. Then, the dimensionality of the load curves was reduced, using the Wavelet Transform, followed by clustering performed with the k-means algorithm. The result of the filtering step demonstrated the effectiveness in correcting abnormalities present in the load curves, thus excluding the load curves with a predominance of abnormalities indicating failures in the measurement system. The clusters resulting from the dimensionality reduction method followed by clustering, originated load curves with well-defined typologies, as well as typical days and periods of the year, in addition to a reduced processing time. The entire methodology was applied using real data from a total of 2588 load curves discretized every 5 minutes. [ABSTRACT FROM AUTHOR]