School of Computer Science and Informatics

Permanent URI for this collection


Recent Submissions

Now showing 1 - 20 of 3494
  • ItemOpen Access
    Risks and benefits of smart toilets
    (ACM, 2023-11-15) Wagner, Isabel; Boiten, Eerke
    Smart toilets promise convenient 24/7 health and wellness monitoring. However, privacy risks of smart toilets have not been carefully studied. Here, we present a thematic analysis of an expert focus group on smart toilets that record health data. The themes indicate severe privacy and systemic risks, many of which could be mitigated but currently are not. Our analysis suggests that health benefits outweigh risks only in specific application contexts.
  • ItemOpen Access
    Network Intrusion Detection based on Amino Acid Sequence Structure Using Machine Learning
    (MDPI, 2023-10-17) Ibaisi, Thaer AL; Kuhn, Stefan; Kaiiali, Mustafa; Kazim, Muhammad
    The detection of intrusions in computer networks, known as Network-Intrusion-Detection Systems (NIDSs), is a critical field in network security. Researchers have explored various methods to design NIDSs with improved accuracy, prevention measures, and faster anomaly identification. Safeguarding computer systems by quickly identifying external intruders is crucial for seamless business continuity and data protection. Recently, bioinformatics techniques have been adopted in NIDSs’ design, enhancing their capabilities and strengthening network security. Moreover, researchers in computer science have found inspiration in molecular biology’s survival mechanisms. These nature-designed mechanisms offer promising solutions for network security challenges, outperforming traditional techniques and leading to better results. Integrating these nature-inspired approaches not only enriches computer science, but also enhances network security by leveraging the wisdom of nature’s evolution. As a result, we have proposed a novel Amino-acid-encoding mechanism that is bio-inspired, utilizing essential Amino acids to encode network transactions and generate structural properties from Amino acid sequences. This mechanism offers advantages over other methods in the literature by preserving the original data relationships, achieving high accuracy of up to 99%, transforming original features into a fixed number of numerical features using bio-inspired mechanisms, and employing deep machine learning methods to generate a trained model capable of efficiently detecting network attack transactions in real-time.
  • ItemMetadata only
    A Multispectral Image Classification Framework for Estimating the Operational Risk of Lethal Wilt in Oil Palm Crops
    (Springer Nature, 2023-04-09) Pena, Alejandro; Puerta, Alejandro; Bonet, Isis; Caraffini, Fabio; Ochoa, Ivan; Gongora, Mario Augusto
    Operational risk is the risk associated with business operations in an organisation. With respect to agricultural crops, in particular, operational risk is a fundamental concept to establish differentiated coverage and to seek protection against different risks. For cultivation, these risks are related to the agricultural business process and to external risk events. An operational risk assessment allows one to identify the limits of environmental and financial sustainability. Specifically, in oil palm cultivation, the characterisation of the associated risk remains a challenge from a technological perspective. To advance in this direction, researchers have used different technologies, including spectral aerial images, unmanned aerial vehicles to construct a vegetation index, intelligent augmented platforms for real-time monitoring, and adaptive fuzzy models to estimate operational risk. In line with these technological developments, in this article we propose a framework for the estimation of the risk assessment associated with the disease of Lethal Wilt (LW) in oil palm plantations. Although our purpose is not to predict lethal wilt, since the framework starts from the result of a prediction model, a model to detect LW in an early stage is used for the demonstration. For the implementation of the prediction model, we use a novel deep learning system based on two neural networks. This refers to a case study conducted at UNIPALMAS. We show that the suitability of our system aims to evaluate operational risks of LW with a confidence level of 99.9% and for a period of 6 months.
  • ItemEmbargo
    A minimum cost-maximum consensus jointly driven feedback mechanism under harmonious structure in social network group decision making
    (Elsevier, 2023-10-30) Wang, Sha; Chiclana, Francisco; Chang, Jia Li; Xing, Yumei; Wu, Jian
    This article investigates a minimum cost-maximum consensus jointly driven feedback mechanism under a harmonious power structure by twofold group and individual attention recommendations for building social network consensus. Harmonious power structure is first constructed with subgroup-centrality-IOWA operator by (i) extracting subgroup importance rankings through social network analysis, and (ii) minimising group structure conflict to search the harmony weight allocation. Subsequently, a twofold attention recommendation approach that balances group attention and individual attention is proposed to generate feedback recommendations for the feedback recipients. Based on this, optimisation models that minimise individual adjustment cost and maximise group consensus are constructed, jointly driving the feedback mechanism. Finally, a demonstration example is provided to illustrate the efficacy of the proposed model.
  • ItemMetadata only
    Motion Pattern-Based Scene Classification Using Adaptive Synthetic Oversampling and Fully Connected Deep Neural Network
    (IEEE, 2023-10-25) Mohammed, Sultan Mohammed; Al-Dhamari, Ahlam; Saeed, Waddah; Al-Aswadi, Fatima N.; Saleh, Sami Abdulla Mohsen; Marsono, M. N.
    Analyzing crowded environments has become an increasingly researched topic, largely due to its myriad practical applications, including enhanced video surveillance systems and the estimation of crowd density in specific settings. This paper presents a comprehensive method for progressing the study of crowd dynamics and behavioral analysis, specifically focusing on the classification of movement patterns. We introduce a specialized neural network-based classifier explicitly designed for the accurate categorization of various crowd scenes. This classifier fills a unique niche in the existing literature by offering robust and adaptive classification capabilities. To optimize the performance of our model, we conduct an in-depth analysis of loss functions commonly employed in multi-class classification tasks. Our study encompasses four widely-used loss functions Focal Loss, Huber Loss, Cross-Entropy Loss, and Multi-Margin Loss. Based on empirical findings, we introduce a Joint Loss function that combines the strengths of Cross-Entropy and Multi-Margin Loss, outperforming existing methods across key performance metrics such as accuracy, precision, recall, and F1-score. Furthermore, we address the critical challenge of class imbalance in motion patterns within crowd scenes. To this end, we perform a comprehensive comparative study of two leading oversampling techniques: the synthetic minority oversampling technique (SMOTE) and adaptive synthetic sampling (ADASYN). Our results indicate that ADASYN is superior at enhancing classification performance. This approach not only mitigates the issue of class imbalance but also provides robust empirical validation for our proposed method. Finally, we subject our model to a rigorous evaluation using the Collective Motion Database, facilitating a comprehensive comparison with existing state-of-the-art techniques. This evaluation confirms the effectiveness of our model and aligns it with established paradigms in the field.
  • ItemEmbargo
    Combining state detection with knowledge transfer for constrained multi-objective optimization
    (IEEE, 0202-04-18) Zheng, Jinhua; Yang, Kaixi; Zou, Juan; Yang, Shengxiang
    The main challenge in studying constrained multi-objective optimization problems (CMOPs) is reasonably balancing convergence, diversity, and feasibility. One of the most successful solutions to this challenge is the co-evolutionary frame-work, an algorithm in which multiple populations cooperate and complement each other, with different but interdependent populations addressing different but related problems. However, the effectiveness of existing algorithms for information exchange between various populations is not apparent. This paper proposes a new algorithm named SDKT using population state detection and knowledge transfer. The method has dual stages (i.e., knowledge acquisition and knowledge reception) and dual populations. Specifically, by restarting the strategy, these two populations (i.e., mainPop and auxPop) first explore more feasible regions with and without constraints. Then, in the knowledge receiving stage, ma i nPop and auxPop provide effective information to promote each other's approach to constrained PF (CPF) and unconstrained PF (UPF), respectively. Extensive experiments on three well-known test suites and three real-world problem studies fully demonstrate that SDKT is more competitive than five state-of-the-art constrained multi-objective evolutionary algorithms.
  • ItemOpen Access
    Ordering vs. AHP. Does the intensity used in the decision support techniques compensate?
    (Elsevier, 2023-10-02) Sáenz-Royo, Carlos; Chiclana, Francisco; Herrera-Viedma, Enrique
    The manifestation of the intensity in the judgment of one alternative versus another in the peer comparison processes is a central element in some decision support techniques, such as the Analytical Hierarchy Process (AHP). However, his contribution in terms of quality (expected performance) with respect to the priority vector has not been evaluated so far. In this work, through the Intentional Bounded Rationality Methodology (IBRM) of Sáenz-Royo, Chiclana, and Herrera-Viedma (2023), the gains obtained from requiring the decision-maker to report an intensity judgment in pairs (AHP) are analyzed with respect to a technique that only requires expressing a preference (Ordering). The results show that when decision-makers have low levels of expertise, it is possible that a less informative and expensive technique (Ordering) performs better than a more informative and expensive one (AHP). When decision-makers have medium and high levels of expertise, AHP obtains meager gains about Ordering. This study proposes a cost-benefit analysis of decision support techniques contrasting the gains of a technique that requires more (AHP) resources with other less expensive (Ordering). Our results can change the way of managing the information obtained from experts’ judgments.
  • ItemMetadata only
    A Framework of Directed Network Based Influence-Trust Fuzzy Group Decision Making
    (Springer Nature, 2023-08-17) Kamis, Nor Hanimah; Kilicman, Adem; Kadir, Norhidayah A.; Chiclana, Francisco
    Daily life requires individuals or groups of decision-makers to engage in critical decision-making processes. Fuzzy set theory has been integrated into group decision-making (GDM) to address the ambiguity and vagueness of expert preferences. Social Network Group Decision Making (SNGDM) is a newly emerging research area that focuses on the use of social networks to facilitate information exchange and communication among experts in GDM. Moreover, Social Influence Group Decision Making (SIGDM) has been initiated, which considers social influence as a factor that can impact experts’ preferences during interactions or discussions. Studies in this area have proposed innovative measurements of social influence, including the use of trust statements to explicitly influence experts’ opinions. In this study, a new trust index called TrustRank is proposed, which acts as an additional weightage of experts’ importance and is embedded in the influence network measure that represents the strength of the expert’s influence degree. These values are then utilized as the order-inducing variable in the IOWA-based fusion operator to obtain the collective preference and ranking of alternatives. The proposed framework, which is the directed network-based Influence-Trust Fuzzy GDM, is presented along with its implementation, results, and discussion to showcase its applicability.
  • ItemEmbargo
    Dynamic niching particle swarm optimization with an external archive-guided mechanism for multimodal multi-objective optimization
    (Elsevier, 2023-10-19) Sun, Yu; Chang, Yuqing; Yang, Shengxiang; Wang, Fuli
    Multimodal multi-objective optimization problems (MMOPs) contain multiple equivalent Pareto optimal sets (PSs) corresponding to the same Pareto front (PF). However, simultaneously locating well-distributed and well-converged multiple equivalent global PSs and PF remains challenging. Therefore, this paper proposes dynamic niching particle swarm optimization (PSO) with an external archive-guided (AG) mechanism, termed DNPSO-AG, for solving MMOPs. In DNPSO-AG, a clustering-based dynamic niching technique is integrated with PSO to divide the population into multiple niches. In addition, a leader updating method controls the updating of the leaders. Furthermore, a novel external archive-guided mechanism guides the evolution of multiple niches and enhances the distribution of solutions, which comprises two strategies: the adaptive division of the external archive strategy, which adaptively divides the external archive into multiple sub-archives, and the distance-based sub-archive and niche matching strategy, which assigns sub-archives to multiple niches for maintenance. The experimental results demonstrate that the proposed DNPSO-AG outperforms seven other state-of-the-art competitors on the CEC 2019 MMOP test suite in terms of the inverted generational distance (IGD) and IGD in the decision space (IGDX) metrics, with improvements of 21.3% and 9.1% over the best-performing competitor, respectively.
  • ItemEmbargo
    A reputation-based trust evaluation model in group decision-making framework
    (Elsevier, 2023-10-20) You, Xinli; Hou, Fujun; Chiclana, Francisco
    In group decision-making (GDM) problems, experts need to communicate and adjust their opinions in order to achieve consensus on the final decision-making output. Since experts may have conflicting opinions, trust can be critical and an important reference to use in the decision-making process when some experts are required to modify opinions. Recently, decision-making models based on trust and reputation have been investigated intensively. However, these research works usually rely on the constructed social trust network and honesty and fairness of the trust ratings from experts are taken for granted. The objective of this study is to develop a reputation-based trust model for GDM framework to obtain the trust relationship among experts from their direct interaction and word of mouth. First, the paper defines a trust credibility measure to filter out malicious experts before trust assessment, and designs direct trust feedback based on the interaction quality. Then, based on this direct trust feedback, the global reputation model is proposed according to the synthetical performance of received and provided trust feedback, which encourages long-term good behaviour and guarantees trustworthy communications and interactions. The reputation-based trust and direct trust feedback together build trust relationship among experts. Finally, a simulation experimental analysis of the proposed trust and reputation models is carried out to verify their effectiveness in trust and reputation establishment among the experts, even under the presence of malicious ones.
  • ItemEmbargo
    Knowledge transfer-based multi-factorial evolutionary algorithm for selective maintenance optimization of multi-state complex systems
    (IEEE, 2023-10) Xu, Yue; Pi, Dechang; Yang, Shengxiang; Zio, Enrico
    This paper focuses on multi-task selective maintenance for multi-state complex systems, with the goal of selecting subsets of feasible maintenance actions on multi-task systems simultaneously due to limited resources. For each task, system characteristic comprises of various configurations such as series, parallel, bridge, and complex, weibull distribution, and multiple states; maintenance characteristic includes perfect maintenance, imperfect maintenance, and minimal repair. Considering these realistic issues, this paper introduces a reliability evaluation approach, including Markov chain, universal generating function, and imperfect maintenance age reduction model. The challenge of solving such kind of problems lies not only in the reliability estimation, but also in the solution method. Since it is the first time to solve the multi-task selective maintenance problem, this paper tailors a novel multi-factorial evolutionary algorithm, with an improved associate mating. In our algorithm, a similarity-based task selection mechanism tries to determine the intensity between inter-task self-evolution and inter-task knowledge transfer, based on the relatedness between tasks; a feedback-based task transfer mechanism adjusts the transfer intensity, with regard to convergence and diversity. Numerical experiments verify the effectiveness of the proposed method compared with the original one.
  • ItemEmbargo
    Weak relationship indicator-based evolutionary algorithm for multimodal multi-objective optimization
    (Elsevier, 2023-10-05) Xiang, Yi; Zheng, Jinhua; Hu, Yaru; Liu, Yuan; Zou, Juan; Deng, Qi; Yang, Shengxiang
    Multimodal multi-objective problems (MMOPs) have multiple equivalent Pareto sets (PSs) that map to the same Pareto optimal front (PF). Traditional multimodal multiobjective algorithms (MMEAs) use strong relationships to guide population convergence, but this can lead to two problems: the population may explore easier-to-search PSs and lose more difficult-to-search PSs, and it may not retain local PSs well. To address these issues, we propose a weak relationship indicator-based MMEA that includes weak convergence indicators and density evaluation indicators. The weak convergence indicator considers the relationship between an individual and its neighbors, while the density evaluation indicator considers the density information of the individual and its neighbors. This allows the population to retain solutions from different PSs during exploration. An archive based on weak convergence indicators also retains excellent solutions generated during the evolution of the population. Experimental results show that our algorithm ranked first in terms of overall score when compared with seven state-of-the-art algorithms using the Friedman Test.
  • ItemEmbargo
    Online sparse representation clustering for evolving data streams
    (IEEE Press, 2023-10) Chen, Jie; Yang, Shengxiang; Fahy, Conor; Wang, Zhu; Guo, Yinan; Chen, Yingke
    Data stream clustering can be performed to discover the patterns underlying continuously arriving sequences of data. A number of data stream clustering algorithms for finding clusters in arbitrary shapes and handling outliers, such as density-based clustering algorithms, have been proposed. However, these algorithms are often limited in their ability to construct and merge microclusters by measuring the Euclidean distances between high-dimensional data objects, e.g., transferring valuable knowledge from historical landmark windows to the current landmark window, and exploiting evolving subspace structures adaptively. We propose an online sparse representation clustering (OSRC) method to learn an affinity matrix for evaluating the relationships among high-dimensional data objects in evolving data streams. We first introduce a low-dimensional projection into sparse representation to adaptively reduce the potential negative influence associated with the noise and redundancy contained in high-dimensional data. Then, we take advantage of the l2,1-norm optimization technique to choose the appropriate number of representative data objects and form a specific dictionary for sparse representation. The specific dictionary is integrated into sparse representation to adaptively exploit the evolving subspace structures of the high-dimensional data objects. Moreover, the data object representatives from the current landmark window can transfer valuable knowledge to the next landmark window. The experimental results based on a synthetic dataset and six benchmark datasets validate the effectiveness of the proposed method compared to that of state-of-the-art methods for data stream clustering.
  • ItemEmbargo
    A reinforcement learning based dynamic multi-objective constrained evolutionary algorithm for open-pit mine truck scheduling
    (IEEE, 2023-09) Qiu, Junxiang; Li, Changhe; Yang, Shengxiang
    Aiming at the truck scheduling problem in the open-pit mine scenario, a truck scheduling model based on real-time ore blending is established, and an adaptive evolution algorithm for truck scheduling based on DCNSGA-III is proposed. In the established scheduling model, the real-time grade variance of the crushing plant is minimized as one of the optimization objectives, and the Q-learning algorithm is introduced to adaptively select one of the most effective operators during the search process. Experiments show that the proposed method can effectively control the grade fluctuation of the ore flow and better scheduling schemes are obtained in comparison with algorithms equipped with the traditional search operator selection methods.
  • ItemEmbargo
    A reinforcement learning-based multi-objective optimization in an interval and dynamic environment
    (Elsevier, 2023-09-22) Xu, Yue; Song, Yuxuan; Pi, Dechang; Chen, Yang; Qin, Shou; Zhang, Xiaoge; Yang, Shengxiang
    There are many fields involving multi-objective optimization problems in presence of dynamic and interval environments (DI-MOPs), in which the number of objective functions is greater than one, the objectives are conflicting with each other, the problem is varying with time, and the parameters are interval-valued. Conflicts between multiple objectives make interval problems more difficult to be optimized in the dynamic environment. Recent works suffer from the lack of accuracy in change severity detection, the lack of adaptability in change response, and insufficient consideration of reducing imprecision. To tackle these issues, a novel reinforcement learning-based algorithm is proposed in this study, which has three original contributions: (1) Internal interval similarity is specially designed for the interval detection of change severity. To be specific, this operator is proposed for higher accuracy, including the hybridization between the interval similarity and point similarity, and the decision of the detection object and strategy. (2) Q learning is embedded into the optimization algorithm to select the optimal change response after the change occurs. The benefit of this operator is that the response mechanism is dynamically changed in accordance to the environments. (3) To reduce the uncertainty of problems, a new crowding distance operator is presented to guide the search to simultaneously increase diversity, speed up convergence, and decrease imprecision. The computational results from the benchmark sets demonstrate that the proposed algorithm is more efficient than other state-of-the-art algorithms, generating Pareto sets with stronger convergence, wider distribution, and less uncertainty.
  • ItemMetadata only
    Recursive Multi-step Time-Series Forecasting for Residual-Feedback Artificial Neural Networks: A Survey
    (Springer, 2023-09-29) Saeed, Waddah; Ghazali, Rozaida
    Residual-feedback artificial neural networks are a type of artificial neural network (ANNs) that have shown better forecasting performance on some time series. One of the challenges of residual-feedback ANNs is by utilizing the previous time step’s observed value, they are only capable of predicting one step ahead in advance. Therefore, it would not be possible to apply them directly in a recursive multi-step forecast strategy. To shed light on this challenge, a systematic literature review was conducted in this paper to find answers to the following three research questions: What are the main motivations behind introducing residual feedback to ANNs? How good are the existing residual-feedback ANNs compared to other forecasting methods in terms of forecasting performance? And what are the existing solutions for recursive multi-step time series forecasting using residual-feedback ANNs? An analysis of 19 studies was conducted to answer these questions. Furthermore, several potential solutions that can be further practically explored are suggested in an attempt to overcome this challenge.
  • ItemMetadata only
    Temporal modelling of long-term heavy metal concentrations in aquatic ecosystems
    (IWA Publishing, 2023-06-02) Bushra, Basmah; Bazneh, Leyla; Deka, Lipika; Wood, Paul J.; McGowan, Suzanne; Das, Diganta B.
    This paper examines a series of connected and isolated lakes in the UK as a model system with historic episodes of heavy metal contamination. A 9-year hydrometeorological dataset for the sites was identified to analyse the legacy of heavy metal concentrations within the selected lakes based on physico-chemical and hydrometeorological parameters, and a comparison of the complementary methods of multiple regression, time series analysis, and artificial neural network (ANN). The results highlight the importance of the quality of historic datasets without which analyses such as those presented in this research paper cannot be undertaken. The results also indicate that the ANNs developed were more realistic than the other methodologies (regression and time series analysis) considered. The ANNs provided a higher correlation coefficient and a lower mean squared error when compared to the regression models. However, quality assurance and pre-processing of the data were challenging and were addressed by transforming the relevant dataset and interpolating the missing values. The selection and application of the most appropriate temporal modelling technique, which relies on the quality of available dataset, is crucial for the management of legacy contaminated sites to guide successful mitigation measures to avoid significant environmental and human health implications.
  • ItemMetadata only
    Artificial Intelligence–Based Ethical Hacking for Health Information Systems: Simulation Study
    (JMIR Publications, 2023-04-25) He, Ying; Zamani, Efpraxia; Yevseyeva, Iryna; Luo, Cunjin
  • ItemEmbargo
    Politics and policy of Artificial Intelligence
    (Wiley, 2023-09-04) Ulnicane, Inga; Erkkilä, Tero
    While recent discussions about Artificial Intelligence (AI) as one of the most powerful technologies of our times tend to portray it as a predominantly technical issue, it also has major social, political and cultural implications. So far these have been mostly studied from ethical, legal and economic perspectives, while politics and policy have received less attention. To address this gap, this special issue brings together nine research articles to advance the studies of politics and policy of AI by identifying emerging themes and setting out future research agenda. Diverse but complementary contributions in this special issue speak to five overarching themes: understanding the AI as co-shaped by technology and politics; highlighting the role of ideas in AI politics and policy; examining the distribution of power; interrogating the relationship between novel technology and continuity in politics and policy; and exploring interactions among developments at local, national, regional and global levels. This special issue demonstrates that AI policy is not an apolitical field that can be dealt with just by relying on knowledge and expertise but requires an open debate among alternative views, ideas, values and interests.
  • ItemOpen Access
    Prediction method of truck travel time in open pit mines based on LSTM model
    (IEEE Press, 2023-07) Ao, Mengting; Li, Changhe; Yang, Shengxiang
    Aiming at the prediction of truck travel time in open pit mines, we established a prediction model based on long short-term memory(LSTM). This model fully accounts for 11 factors, including the nature of trucks, weather, road conditions, and driver's behaviors, as well as the influence of neighbor road segments in the route on the current predicted road segment. The experiment shows that the error of the LSTM prediction model is significantly reduced compared with SVR and BP models. In addition, the maximum absolute mean error under different conditions is less than 12 seconds.