Showing papers in "International Journal of Computational Intelligence Studies in 2017"

Application of cuckoo search in water quality prediction using artificial neural network

Abstract: Domestic and industrial pollution affected the water quality to a greater extent. Recent research studies have achieved reasonable success in predicting the water quality using several machine learning based techniques. In the current work, a proposed cuckoo search (CS) has been applied to improve the support in the classification process during the water quality prediction. The proposed model (NN-CS) gradually minimises an objective function; namely the root mean square error (RMSE) in order to find the optimal weight vector for the artificial neural network (ANN). The proposed model was compared with three other well-established models, namely NN-GA (ANN trained with genetic algorithm) and NN-PSO (ANN trained with particle swarm optimisation) in terms of accuracy, precision, recall, f-measure, Matthews correlation coefficient (MCC) and Fowlkes-Mallows index (FM index). The simulation results established superior accuracy of NN-CS over the other models.

Intelligent information system to ensure quality in higher education institutions, towards an automated e-university

Abstract: Despite great efforts to assure quality in higher education institutions, the ambiguity of its related concepts and requirements constitute a big challenge when trying to implement it as an intelligent information system. The present work introduces a framework for an intelligent information system that manages the quality assurance in higher education's institutions. The aim of designing such a system is to provide an automation tool that avoids unnecessary and redundant tasks associated to quality in higher education institutions. In addition, the proposed system helps all higher education stockholders to handle and monitor their tasks. Moreover, it aims to help the quality assurance units in a higher education institution to apply their quality's standards and to make sure that they are being maintained and enhanced. This information system contains a core module and 18 sub-modules, which are described in detail. Finally, the characteristics and components of each of these sub-modules are also discussed.

Intelligent systems based on loud computing for healthcare services: a survey

Abstract: Cloud computing plays an important role in healthcare services (HCS) due to its the ability to retrieve patients' data, diagnosis of diseases and other medical fields in less time and less cost. This paper presents a survey of intelligent systems based on cloud environment for HCS. It reviews the uses of intelligent techniques such as genetic algorithm (GA), particle swarm optimisation (PSO) and parallel particle swarm optimisation (PPSO) on cloud computing environment to enhance task scheduling, reduce execution time of requests from stakeholders (patients, doctors, nurses, e.g.) and maximise of resources utilisation on clouds. This paper presents evaluation criteria of selected researches based on accuracy, usability, agility and applied method. Selected researches in this field were reviewed, analysed, summarised and compared according to the used intelligent techniques in cloud computing, healthcare systems and the concluded results. This paper also proposes architecture for intelligent healthcare systems based on cloud computing environment.

SVM-based sentiment classification: a comparative study against state-of-the-art classifiers

Abstract: Transforming the unstructured textual information contained in various social media streams into useful business knowledge is an extremely difficult computational task, mainly, due to the underlying hard pattern classification problem of sentiment analysis, especially within the context of the Greek language. In this paper, we address the pattern classification problem of sentiment analysis through the utilisation of support vector machines (SVMs). In particular, we conducted an extensive experimental comparison where we tested the aforementioned classifier against a set of state-of-the-art machine learning classifiers on a benchmark dataset originating from the Greek bank sector by collecting data from the streaming API of Twitter that were explicitly referring to the major banks of Greece. Our results present classification accuracy and execution time metrics for each classifier, revealing the superiority of the SVM learning paradigm in assigning patterns to the correct sentiment class.

Fine tuning of adaptive learning of deep belief network for misclassification and its knowledge acquisition

Abstract: We have proposed an adaptive structure learning of deep belief network (DBN) that can determine the suitable number of hidden layers and hidden neurons of restricted Boltzmann machines (RBMs). The method shows high classification performance to the big data benchmark test. However, the method could not classify the unknown pattern correctly, since an input data with ambiguous patterns leads the classification to the wrong judgment. In such a case, a fine-tuning method that patches a part of network signal flow based on the knowledge will be a helpful method even in terms of both the improvement of classification capability and the reduction of computational cost by learning again. In this paper, network signal patterns which lead the given misclassified patterns were visualised for knowledge acquisition. By fine-tuning the trained network using the acquired knowledge, the classification capability can achieve great success.

A secure ID-based signcryption scheme based on elliptic curve cryptography

Abstract: Signcryption is a cryptographic primitive which at the same time give both the capacity of digital signature and public key encryption in a single logical step. Identity based cryptography is a distinct option for the traditional certificate based cryptosystem. Its principal thought is that every client utilises his identity information as his public key. Elliptic curve cryptosystem (ECC) have new received consistent attention because of their higher security per bit as compare to other cryptosystem. This paper presents a new identity based signcryption based on elliptic curve cryptography. Its security is dependent on elliptic curve discrete logarithm problem (ECDLP) and elliptic curve Diffie-Hellman problem (ECDHP). The proposed scheme can be very useful in low-end resource devices such as mobile communication, mobile banking, personal digital assistant (PDA) and internet of things (IoT).

Modelling of input-output relationships of metal inert gas welding process using soft computing-based approaches

Abstract: Input-output relationships of metal inert gas welding process were determined in both forward and reverse directions, which are required in order to automate the same. Various types of neural networks, namely multi-layer feed-forward network, counter-propagation network and radial basis function network had been used for the said purpose. The networks were trained using back-propagation algorithm and/or genetic algorithm. Their performances were compared, and radial basis function network developed using the concept of clustering was found to perform better than other networks in terms of accuracy in prediction.

Intelligent comfort management agent for smart residential buildings using an updated Q learning algorithm

Abstract: Development of smart environments is one of the hot researching fields of this digital era. The goal of the presented work is to investigate the applicability of reinforcement learning technique for designing intelligent comfort management systems of smart residences which considers minimising the electricity consumption as its hidden agenda while maintaining maximum comfort of the occupants. Accurate occupancy estimation of a smart homes equipped with ambient sensing is expected to give vital inputs to intelligent appliance scheduling algorithms. The proposed Q learning based intelligent comfort management agent (Q-ICMA) dynamically estimates the occupancy level of the given smart space through ambient sensors embedded in the environment and then utilises this information to drive the environment to the optimum region by automatically controlling the lighting and ventilation systems using Q learning algorithm. Simulation results show that the e updated Q learning based agent achieves the best possible results in terms of maximum rewards and faster convergence in achieving the desired goal state.

Computational intelligence in telecommunication networks: a review

Abstract: The recent increase in the demand for process optimisation, something which we continually learn from nature, help the telecommunication arena to rapidly evolve into a more scalable, dynamic and robust systematic structure. Due to high intricacy and complex dynamism in every consortium of communication topology, especially in this era of big data and IoT, traditional methods fail to provide network performance and desired throughputs accurately. Computational intelligence now-a-days become indispensable for telecommunication due to the tremendous complexity in multidimensional work space and dynamic environment. Advanced high-tech machines should have their own bio-inspired adaptive intelligence with which they will emerge as energy-efficient and sustainable alternatives for our future generations. In this paper, we try to review different evolving intelligent techniques those are implemented in variegated dimensions of optimisations in telecommunication networks.

An FCA approach to mining quantitative association rules from multi-relational data

Abstract: In this paper, we propose an algorithm for mining quantitative association rules (ARs) from a multi-relational database (MRDB) A MRDB contains multiple tables (relations), and attributes in a table are either categorical or quantitative (or numerical) To handle numerical data in a pattern, we consider (logical) conjunctions with interval constraints, using the notion of closed interval patterns (CIPs) proposed by Kaytoue et al in formal concept analysis (FCA) We then present an algorithm for mining strong quantitative ARs, namely they satisfy both a minimum support and a minimum confidence We also propose a pruning method tailored to computing CIPs in an AR We give some experimental results, which show the effectiveness of the proposed method, compared with the conventional methods such as a discretisation-based approach or an optimisation-based approach