Search

Image recaptured from a high-resolution LED screen or a good quality printer is difficult to distinguish from its original counterpart. The forensic community paid less attention to this type of forgery than to other image alterations such as splicing, copy-move, removal, or image retouching. It is significant to develop secure and automatic techniques to distinguish real and recaptured images without prior knowledge. Image manipulation traces can be hidden using recaptured images. For this reason, being able to detect recapture images becomes a hot research topic for a forensic analyst. The attacker can recapture the manipulated images to fool image forensic system. As far as we know, there is no prior research that has examined the pros and cons of up-to-date image recaptured techniques. The main objective of this survey was to succinctly review the recent outcomes in the field of image recaptured detection and investigated the limitations in existing approaches and datasets. The outcome of this study provides several promising directions for further significant research on image recaptured detection. Finally, some of the challenges in the existing datasets and numerous promising directions on recaptured image detection are proposed to demonstrate how these difficulties might be carried into promising directions for future research. We also discussed the existing image recaptured datasets, their limitations, and dataset collection challenges.

Dynamic traffic in a software-defined network (SDN) causes explosive data to flow from one system to another. The explosive data affects the functionality of system parameters, network-level configuration, routing parameters, network characteristics, and system load factors. Adapting to the traffic flow is a key research area in SDN in today’s big data world. Load balance vehicular sensor accessibility reduces delays, lowers energy consumption, and decreases the execution time. This paper combines the entropy-based active learning model to identify intrusion patterns efficiently, which is a packet-level intrusion detection model. The developed afterload balancing model can track the attack on the network. We then proposed a load balancing algorithm that optimizes the vehicular sensor usability by using sensor computing capability and source needs. We make use of a convergence-based mechanism to achieve high resource utilization. We then perform experiments on the state-of-the-art intrusion detection dataset. Our experimental results show that the load balancing mechanism can achieve 2x in performance improvements compared to traditional approaches. Thus, we can see that the designed model can help improve the decision boundary by increasing the training instance through pooling strategy and entropy uncertainty measure.

Artificial Intelligence (AI) is increasingly becoming an important component of the education sector, both in primary education and in higher education. The rapid advancement of computing technologies and big data processing techniques has made it possible to implement AI in education. AI and machine learning techniques are now being used to deliver chatbots like ChatGPT; intelligent tutoring systems; automatic grading systems; and other advanced learning systems like AI Adaptive Learning Systems, in the education sector. These advanced learning systems, AI enabled Adaptive Learning Systems (AI-ALS), is the focus of my doctoral dissertation. AI-ALS are the most advanced generation of adaptive learning systems that use more sophisticated data analytics (learning analytics) and AI to provide just-in-time insights and embed ongoing data collection to improve the quality of adaptive learning. Adaptive learning is the personalization of learning for students in a learning system. Thus, AI-ALS are learning platforms that adapt to the learning strategies of students by modifying the order and difficulty level of learning tasks. Adaptive learning systems, which personalize the learning experience to the needs and abilities of students, are a promising application of artificial intelligence in education. A growing body of research suggests that the use of AI-ALS in educational contexts has the potential to enhance teaching and learning skills as well as human performance. However, the effectiveness of these systems in real-world settings is not yet clear. While the potential and importance of such systems are well documented, the actual implementation status of AI-ALS and other AI-based learning systems in real-life teaching and learning settings is low and not fully understood. Also, the research on AI-ALS lacks insights and codification of its design and development knowledge. In addition, misuse of AI due to algorithm biases and lack of governance could also exacerbate inequalities in our educational institutions.

During crisis situations, emergency operators require fast information access to achieve situation awareness and make the best possible decisions. Augmented reality could be used to visualize the wealth of user-generated content available on social media and enable context-adaptive functions for emergency operators. Although emergency operators agree that social media analytics will be important for their future work, it poses a challenge to filter and visualize large amounts of social media data. We conducted a goal-directed task analysis to identify the situation awareness requirements of emergency operators. By collecting tweets during two storms in Germany we evaluated the usefulness of Twitter data for achieving situation awareness and conducted interviews with emergency operators to derive filter strategies for social media data. We synthesized the results by discussing how the unique interface of augmented reality can be used to integrate social media data into emergency control rooms for situation awareness support.

This paper investigates the attitude tracking control problem for uncertain nonlinear rigid body systems, where both inputs and states are quantized. It is common in networked control systems that sensor and control signals are quantized before they are transmitted via a communication network. An adaptive backstepping control algorithm is developed for a class of uncertain multiple-input multiple-output (MIMO) systems where a class of sector bounded quantizers is considered. It is shown that all the closed-loop signals are ensured uniformly bounded and tracking is achieved. Further, the tracking errors are shown to converge towards a compact set containing the origin and the set can be made small by the choice of the quantization parameters and the control parameters. For illustration of the proposed control scheme, experiments were conducted on a 2 degrees-of-freedom (DOF) helicopter system.

The present era is facing the industrial revolution. Machine-to-Machine (M2M) communication paradigm is becoming prevalent. Resultantly, the computational capabilities are being embedded in everyday objects called things. When connected to the internet, these things create an Internet of Things (IoT). However, the things are resource-constrained devices that have limited computational power. The connectivity of the things with the internet raises the challenges of the security. The user sensitive information processed by the things is also susceptible to the trusability issues. Therefore, the proliferation of cybersecurity risks and malware threat increases the need for enhanced security integration. This demands augmenting the things with state-of-the-art deep learning models for enhanced detection and protection of the user data. Existingly, the deep learning solutions are overly complex, and often overfitted for the given problem. In this research, our primary objective is to investigate a lightweight deep-learning approach maximizes the accuracy scores with lower computational costs to ensure the applicability of real-time malware monitoring in constrained IoT devices. We used state-of-the-art Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), and Bi-directional LSTM deep learning algorithm on a vanilla configuration trained on a standard malware dataset. The results of the proposed approach show that the simple deep neural models having single dense layer and a few hundred trainable parameters can eliminate the model overfitting and achieve up to 99.45% accuracy, outperforming the overly complex deep learning models.

Healthcare sector is one of the prominent sectors in which a lot of data can be collected not only in terms of health but also in terms of finances. Major frauds happen in the healthcare sector due to the utilization of credit cards as the continuous enhancement of electronic payments, and credit card fraud monitoring has been a challenge in terms of financial condition to the different service providers. Hence, continuous enhancement is necessary for the system for detecting frauds. Various fraud scenarios happen continuously, which has a massive impact on financial losses. Many technologies such as phishing or virus-like Trojans are mostly used to collect sensitive information about credit cards and their owner details. Therefore, efficient technology should be there for identifying the different types of fraudulent conduct in credit cards. In this paper, various machine learning and deep learning approaches are used for detecting frauds in credit cards and different algorithms such as Naive Bayes, Logistic Regression, K-Nearest Neighbor (KNN), Random Forest, and the Sequential Convolutional Neural Network are skewed for training the other standard and abnormal features of transactions for detecting the frauds in credit cards. For evaluating the accuracy of the model, publicly available data are used. The different algorithm results visualized the accuracy as 96.1%, 94.8%, 95.89%, 97.58%, and 92.3%, corresponding to various methodologies such as Naive Bayes, Logistic Regression, K-Nearest Neighbor (KNN), Random Forest, and the Sequential Convolutional Neural Network, respectively. The comparative analysis visualized that the KNN algorithm generates better results than other approaches.

The elegance of the single-pushout (SPO) approach to graph transformations arises from substituting total morphisms by partial ones in the underlying category. SPO's applicability depends on the durability of pushouts after this transition. There is a wide range of work on the question when pushouts exist in categories with partial morphisms starting with the pioneering work of Lowe and Kennaway and ending with an essential characterisation in terms of an exactness property (for the interplay between pullbacks and pushouts) and an adjointness condition (w.r.t. inverse image functions) by Hayman and Heindel. Triple graphs and graph diagrams are frameworks to synchronise two or more updatable data sources by means of internal mappings, which identify common sub-structures. Comprehensive systems generalise these frameworks, treating the network of data sources and their structural inter-relations as a homogeneous comprehensive artefact, in which partial maps identify commonalities. Although this inherent partiality produces amplified complexity, we can show that Heindel's characterisation still yields existence of pushouts in the category of comprehensive systems and reflective partial morphisms and thus enables computing by typed SPO graph transformation.

This is the author's version of the work. It is posted here for your personal use. Not for redistribution. The definitive Version of Record was published in ACM Transactions on Internet Technology, Volume 21, Issue 4, https://doi.org/10.1145/3423557. In the ever-growing world, the concepts of High-utility Itemset Mining (HUIM) as well as Frequent Itemset Mining (FIM) are fundamental works in knowledge discovery. Several algorithms have been designed successfully. However, these algorithms only used one factor to estimate an itemset. In the past, skyline pattern mining by considering both aspects of frequency and utility has been extensively discussed. In most cases, however, people tend to focus on purchase quantities of itemsets rather than frequencies. In this article, we propose a new knowledge called skyline quantity-utility pattern (SQUP) to provide better estimations in the decision-making process by considering quantity and utility together. Two algorithms, respectively, called SQU-Miner and SKYQUP are presented to efficiently mine the set of SQUPs. Moreover, the usage of volunteer computing is proposed to show the potential in real supermarket applications. Two new efficient utility-max structures are also mentioned for the reduction of the candidate itemsets, respectively, utilized in SQU-Miner and SKYQUP. These two new utility-max structures are used to store the upper-bound of utility for itemsets under the quantity constraint instead of frequency constraint, and the second proposed utility-max structure moreover applies a recursive updated process to further obtain strict upper-bound of utility. Our in-depth experimental results prove that SKYQUP has stronger performance when a comparison is made to SQU-Miner in terms of memory usage, runtime, and the number of candidates.

This is a preprint. The version of record of this article, first published in Applied Intelligence, is available online at Publisher’s website: https://doi.org/10.1007/s10489-020-02080-w Traditional association-rule mining (ARM) considers only the frequency of items in a binary database, which provides insufficient knowledge for making efficient decisions and strategies. The mining of useful information from quantitative databases is not a trivial task compared to conventional algorithms in ARM. Fuzzy-set theory was invented to represent a more valuable form of knowledge for human reasoning, which can also be applied and utilized for quantitative databases. Many approaches have adopted fuzzy-set theory to transform the quantitative value into linguistic terms with its corresponding degree based on defined membership functions for the discovery of FFIs, also known as fuzzy frequent itemsets. Only linguistic terms with maximal scalar cardinality are considered in traditional fuzzy frequent itemset mining, but the uncertainty factor is not involved in past approaches. In this paper, an efficient fuzzy mining (EFM) algorithm is presented to quickly discover multiple FFIs from quantitative databases under type-2 fuzzy-set theory. A compressed fuzzy-list (CFL)-structure is developed to maintain complete information for rule generation. Two pruning techniques are developed for reducing the search space and speeding up the mining process. Several experiments are carried out to verify the efficiency and effectiveness of the designed approach in terms of runtime, the number of examined nodes, memory usage, and scalability under different minimum support thresholds and different linguistic terms used in the membership functions.

This article explores the gender digital divide in e-government use in Turkey. Historical trends and differences in internet and e-government use by males and females are investigated. The findings of this study indicate that although internet use has increased over the years, there remains a steady and significant gap between males and females in terms of internet use and e-government services use. It appears that females in Turkey tend to be on the wrong side of the digital divide, and a significant number of females do not benefit from the potential advantages of the internet and e-government services. The gender digital gap in e-government use is higher than in internet use.

Following the growing popularity of social commerce sites, there is an increased interest in understanding how consumers decide what products to purchase based on the available information. Consumers nowadays are confronted with the task of assessing marketer-generated (MGC) as well as user-generated information (UGC) in a range of different forms to make informed purchase-related decisions. This study examines the information types and forms that influence consumers in their decision-making process on social commerce. Building on uses and gratifications and dual-process theories, we distinguish between marketer and user generated content, and differentiate formats into informational and normative. Using a mixed methods approach that builds on an eye-tracking study, followed by semi-structured interviews with 23 participants, our results indicate significant differences in the types and format of information consumed for selected versus eliminated products. Specifically, we looked at engagement, cognitive processing, and observation of consumers, since they reveal information about the mental and processing mechanisms during decision making. We find that consumers present a number of differences in terms of these measures among the different types of content, and with respect to selected versus eliminated products. The outcomes of the interviews also serve to complement these findings, providing more detailed information about the processes and emotional states of consumers during the selection process. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

The goal of this paper is to design a new correlation coefficient for T -spherical fuzzy sets (TSFSs), which can accurately measure the nature of correlation (i.e., positive and negative) as well as the degree of relationship between TSFS. In order to formulate our proposed idea, we had taken inspiration from the statistical concept of the correlation coefficient. While doing so, we firstly introduce the variance and covariance of two TSFS and then constructed our scheme using these two newly defined notions. The numerical value of our proposed correlation coefficient lies within the interval − 1 , + 1 , as it should be from a statistical point of view, whereas the existing methods cannot measure the negative correlation between TSFS, as their numerical value falls within the interval 0 , 1 , which is not reasonable both statistically and intuitively. This aspect has also been thoroughly demonstrated using some numerical examples. The comparison results witnessed the dominance and upper hand of our proposed method over the existing definitions, with reliable and better results. In order to demonstrate the feasibility, usefulness, and practical application, we applied our proposed scheme to solve technical and scientific problems of multicriteria decision-making and pattern recognition. The numerical results show that our proposed scheme is practically suitable, technically applicable, and intuitively reasonable.

Paper A is © IEEE. Papers B and C are reproduced with permission from Springer Nature. In reference to IEEE copyrighted material which is used with permission in this thesis, the IEEE does not endorse any of Western Norway University of Applied Sciences’s products or services. Internal or personal use of this material is permitted. If interested in reprinting/republishing IEEE copyrighted material for advertising or promotional purposes or for creating new collective works for resale or redistribution, please go to http://www.ieee.org/publications_standards/publications/rights/rights_link.html to learn how to obtain a License from RightsLink. To ensure ultra-high dependability and ultra-low defect rates, certification standards such as DO-178C requires safety-critical software with the highest safety level (Level A) in avionics systems to conform to the modified condition decision coverage (MC/DC) criterion. MC/DC is a strong coverage criterion that subsumes existing coverage criteria and it requires a small number of test inputs compared to the combinatorially exhaustive multiple condition coverage (MCC). MC/DC has also proven to reveal many program defects. However, both MC/DC measurement and generating test cases satisfying MC/DC remain a challenging task. In addition, related properties such data races detection can be monitored using some methods used to check MC/DC, as good concurrency coverage increases a likelihood of catching concurrent-related bugs. To address the above challenges, existing strategies rely on intrusive instrumentation which is not recommended for safety critical software since it consumes valuable resources and can alter the behaviour of the system under test (SUT) if it remains in the released code. To overcome the above challenges, this thesis introduces novel paradigms and tools for software testing based on traces and coverage analysis. Our aim is to analyse the MC/DC without instrumentation and to monitor data races with a lightweight instrumentation. In addition, we explore the applicability of MC/DC criterion on the design level models. Furthermore, we investigate new techniques for test cases generation satisfying MC/DC with the aim to increase the coverage. The scientific contribution of this thesis is fourfold: First, we propose an approach for measuring MC/DC without instrumentation. This has resulted in a tooling for MC/DC measurement and analysis based on the trace of an executing program. A static analysis is used to find conditional jumps in object code that correspond to conditions in the source code. With that information the assignments of the conditions during the execution of the code can be reconstructed by analyzing the trace. MC/DC is then evaluated and the covered/uncovered conditionals in the program can be identified. This approach is evaluated on C programs. Secondly, we provide a non-intrusive tooling for data races detection using the continuous observation of embedded multicore systems (COEMS) technology through continuous online monitoring with lightweight instrumentation on a novel FPGA-based external platform for embedded multicore systems. It is used in combination with formal specifications in the high-level temporal stream-based specification language (TeSSLa), in which we encode a lockset-based algorithm to indicate potential race conditions. We show how to instantiate a TeSSLa template that is based on the Eraser algorithm, and present a corresponding light-weight instrumentation mechanism that emits the required observations to the FPGA with low overhead. Thirdly, we investigated the applicability of MC/DC criterion on design level models, where specifically, we conducted a coverage analysis to Coloured Petri Nets (CPNs) models. We implement a library for CPN Tools and a post-processing tool for MC/DC coverage analysis of net inscriptions on a set of model executions and evaluate our approach on eleven larger publicly available CPN models. In the fourth contribution, we propose a new and alternative strategy for test case generation satisfying MC/DC.We have implemented an algorithm for MC/DC test cases based on binary decision diagrams (BDDs) and evaluated on Traffic Alert and Collision Avoidance System (TCAS II) benchmarks. A performance evaluation with respect to the state-of-the art in the form of related work has been conducted.

Explainability is one of the key factors in Natural Language Processing (NLP) specially for legal documents, medical diagnosis, and clinical text. Attention mechanism has been a popular choice for such explainability recently by estimating the relative importance of input units. Recent research has revealed, however, that such processes tend to misidentify irrelevant input units when explaining them. This is due to the fact that language representation layers are initialized by pre-trained word embedding that is not context-dependent. Such a lack of context-dependent knowledge in the initial layer makes it difficult for the model to concentrate on the important aspects of input. Usually, this does not impact the performance of the model, but the explainability differs from human understanding. Hence, in this paper, we propose an ensemble method to use logic-based information from the Tsetlin Machine to embed it into the initial representation layer in the neural network to enhance the model in terms of explainability. We obtain the global clause score for each word in the vocabulary and feed it into the neural network layer as context-dependent information. Our experiments show that the ensemble method enhances the explainability of the attention layer without sacrificing any performance of the model and even outperforming in some datasets.

As a result of several governments closing educational facilities in reaction to the COVID-19 pandemic in 2020, almost 80% of the world’s students were not in school for several weeks. Schools and universities are thus increasing their efforts to leverage educational resources and provide possibilities for remote learning. A variety of educational programs, platforms, and technologies are now accessible to support student learning; while these tools are important for society, they are primarily concerned with the dissemination of theoretical material. There is a lack of support for hands-on laboratory work and practical experience. This is particularly important for all disciplines related to science, technology, engineering, and mathematics (STEM), where labs and pedagogical assets must be continuously enhanced in order to provide effective study programs. In this study, we describe a unique perspective to achieving multi-sensory learning through the integration of virtual and augmented reality (VR/AR) with haptic wearables in STEM education. We address the implications of a novel viewpoint on established pedagogical notions. We want to encourage worldwide efforts to make fully immersive, open, and remote laboratory learning a reality., European Union through the Erasmus+ Program under Grant 2020-1-NO01-KA203-076540, project title Integrating virtual and AUGMENTED reality with WEARable technology into engineering EDUcation (AugmentedWearEdu), https://augmentedwearedu.uia.no/ [34] (accessed on 27 March 2022). This work was also supported by the Top Research Centre Mechatronics (TRCM), University of Agder (UiA), Norway

Identity-based encryption is an important cryptographic system that is employed to ensure confidentiality of a message in communication. This article presents a provably secure identity based encryption based on post quantum security assumption. The security of the proposed encryption is based on the hard problem, namely Learning with Errors on integer lattices. This construction is anonymous and produces pseudo random ciphers. Both public-key size and ciphertext-size have been reduced in the proposed encryption as compared to those for other relevant schemes without compromising the security. Next, we incorporate the constructed identity based encryption (IBE) for Internet of Things (IoT) applications, where the IoT smart devices send securely the sensing data to their nearby gateway nodes(s) with the help of IBE and the gateway node(s) secure aggregate the data from the smart devices by decrypting the messages using the proposed IBE decryption. Later, the gateway nodes will securely send the aggregated data to the cloud server(s) and the Big data analytics is performed on the authenticated data using the Artificial Intelligence (AI)/Machine Learning (ML) algorithms for accurate and better predictions.

Estimating the remaining useful life (RUL) of bearings from healthy to faulty is important for predictive maintenance. The bearing fault severity can be estimated based on the energy or root mean square (RMS) of vibration signals, and a stopping criterion can be set based on a threshold given by an ISO standard. However, the vibration RMS is often not monotonically increasing with damage, which renders a challenge for predicting the RUL. This study proposes a novel method for splitting the vibration signal into multiple frequency bands before RMS calculations to generate multiple health indicators. Monotonic health indicators are identified using the Spearman coefficient, and the RUL is afterward estimated for each indicator using a suitable model and parameter update scheme. Historical failure data is not required to set any parameters. The proposed method is tested with the Paris' law, where parameters are updated by particle filters. Experimental results from two test rigs validate the performance of the proposed method.

Customers’ data points sources are growing, through the Internet of Things (IoT), telematics and interfacing with third parties’ platforms. Third parties provide either by-product data about their clients, or data that are generated for the specific task of a more accurate measurement of behaviour. One of the core building blocks in insurance is understanding risk – therefore, innovation around data points (sources and analytics) to proxy risk is the engine behind its digital transformation. While the value of data points is clear for the providers of insurance products and services, the customers have not only less control over their data, but also hardly benefit from the creation of those data points. In this paper we provide a case study from the British car insurance industry, about a data-driven product that pivoted into a re-designed personal data right architecture, which included transparency, traceability and usability of customers’ digital traces, and enabled customers’ ownership rights over their data.

Software systems have become crucial for society and the economy to function. Constantly they are permeating more and more application domains. Also, they are getting increasingly integrated with already existing systems. As a result, the development and maintenance of such systems are getting increasingly complex as well. Abstraction is a central key to tackle this complexity. Thus, the software engineering research discipline conceives software systems through the means of models, i.e. simplified representations of reality that simultaneously describe and prescribe the structure and the behaviour of these systems. When engineering system’s integration, the overall design involves multiple models at the same time because each model focuses on a specific aspect of the overall system. This setting is commonly called a multi-model. A major open challenge in software engineering is to maintain the consistency of a multi-model: The individual members of a multi-model are often overlapping and are authored by distinct stakeholders, which may result in inconsistencies, which negatively affect the quality of the system under development. Hence, there is a necessity for a recurring activity, which I am going to call multi-model consistency management throughout this thesis. This general activity includes several challenging software engineering sub-problems such as aligning heterogeneous models, automatic consistency verification and model repair, therefore touching on a range of related research fields. In this thesis, I address limitations with existing approaches and develop a solution to the issue of multi-model consistency management. The fundamental novelty of my solution is its ability to handle multi-models of arbitrary arity because established approaches are mainly limited to binary situations. Software engineering is a constructive scientific discipline and, therefore, this thesis follows a constructive approach. In particular, three artefacts are constructed: (i) A conceptual model of the problem domain; (ii) A formalism called comprehensive systems (iii) A prototype tool called CorrLang. Each artefact marks a contribution to the existing body of scientific knowledge providing value for both researchers and practitioners in software engineering. The conceptual model provides methodological value for software engineers, software architects and decision-makers dealing with multi-model consistency management problems by providing guidelines for analysing concrete scenarios. The formalism provides theoretical value and allows to abstractly represent and reason about inter-related software models in a technology-independent way, which can be used to develop solutions on a high level of abstraction. Simultaneously, it represents a generalisation of existing formalisms that have been used to describe the synchronisation of multi-models. Finally, the prototype provides practical value, which on the one hand, demonstrates the implementability of the conceptual and the formal framework, and on the other hand, it is directly applicable for end-users to solve concrete multi-model consistency management problems.

Today, many fields are characterised by having extensive quantities of data from a wide range of dissimilar sources and domains. One such field is medicine, in which data contain exhaustive combinations of spatial, temporal, linear, and relational data. Often lacking expert-assessed labels, much of this data would require analysis within the fields of unsupervised or semi-supervised learning. Thus, reasoned by the notion that higher view-counts provide more ways to recognise commonality across views, contrastive multi-view clustering may be utilised to train a model to suppress redundancy and otherwise medically irrelevant information. Yet, standard multi-view clustering approaches do not account for relational graph data. Recent developments aim to solve this by utilising various graph operations including graph-based attention. And within deep-learning graph-based multi-view clustering on a sole view-invariant affinity graph, representation alignment remains unexplored. We introduce Deep Representation-Aligned Graph Multi-View Clustering (DRAGMVC), a novel attention-based graph multi-view clustering model. Comparing maximal performance, our model surpassed the state-of-the-art in eleven out of twelve metrics on Cora, CiteSeer, and PubMed. The model considers view alignment on a sample-level by employing contrastive loss and relational data through a novel take on graph attention embeddings in which we use a Markov chain prior to increase the receptive field of each layer. For clustering, a graph-induced DDC module is used. GraphSAINT sampling is implemented to control our mini-batch space to capitalise on our Markov prior. Additionally, we present the MIMIC pleural effusion graph multi-modal dataset, consisting of two modalities registering 3520 chest X-ray images along with two static views registered within a one-day time frame: vital signs and lab tests. These making up the, in total, three views of the dataset. We note a significant improvement in terms of separability, view mixing, and clustering performance comparing DRAGMVC to preceding non-graph multi-view clustering models, suggesting a possible, largely unexplored use case of unsupervised graph multi-view clustering on graph-induced, multi-modal, and complex medical data.

Climate change is stated as one of the largest issues of our time, resulting in many unwanted effects on life on earth. Cloud fractional cover (CFC), the portion of the sky covered by clouds, might affect global warming and different other aspects of human society such as agriculture and solar energy production. It is therefore important to improve the projection of future CFC, which is usually projected using numerical climate methods. In this paper, we explore the potential of using machine learning as part of a statistical downscaling framework to project future CFC. We are not aware of any other research that has explored this. We evaluated the potential of two different methods, a convolutional long short-term memory model (ConvLSTM) and a multiple regression equation, to predict CFC from other environmental variables. The predictions were associated with much uncertainty indicating that there might not be much information in the environmental variables used in the study to predict CFC. Overall the regression equation performed the best, but the ConvLSTM was the better performing model along some coastal and mountain areas. All aspects of the research analyses are explained including data preparation, model development, ML training, performance evaluation and visualization.

Virtual reality games are playing a greater role in rehabilitation settings. Previously, commercial games have dominated, but increasingly, bespoke games for specific rehabilitation contexts are emerging. Choice and design of tasks for VR-games are still not always clear, however; some games are designed to motivate and engage players, not necessarily with the facilitation of specific movements as a goal. Other games are designed specifically for the facilitation of specific movements. A theoretical background for the choice of tasks seems warranted. As an example, we use a game that was designed in our lab: VR Walk. Here, the player walks on a treadmill while wearing a head-mounted display showing a custom-made virtual environment. Tasks include walking on a glass bridge across a drop, obstacle avoidance, narrowing path, walking in virtual footsteps, memory, and selection tasks, and throwing and catching objects. Each task is designed according to research and theory from movement science, exercise science, and cognitive science. In this article, we discuss how for example walking across a glass bridge gives perceptual challenges that may be suitable for certain medical conditions, such as hearing loss, when perceptual abilities are strained to compensate for the hearing loss. In another example, walking in virtual footsteps may be seen as a motor and biomechanical constraint, where the double support phase and base of support can be manipulated, making the task beneficial for falls prevention. In a third example, memory and selection tasks may challenge individuals that have cognitive impairments. We posit that these theoretical considerations may be helpful for the choice of tasks and for the design of virtual reality games.

Electricity network modelling and grid simulations form a key enabling element for the integration of newer and cleaner technologies such as renewable energy generation and electric vehicles into the existing grid and energy system infrastructure. This paper reviews the models of the African electricity systems and highlights the gaps in the open model landscape. Using PyPSA (an open Power System Analysis package), the paper outlines the pathway to a fully open model and data to increase the transparency in the African electricity system planning. Optimisation and modelling can reveal viable pathways to a sustainable energy system, aiding strategic planning for upgrades and policy-making for accelerated integration of renewable energy generation and smart grid technologies such as battery storage in Africa.

Energy as a service (EaaS) is an emerging business model that enables the otherwise passive energy consumers to play an active role and participate in the energy utility services. This platform is formed through smart contracts registering peer-to-peer (P2P) transactions of energy through price and quantity. Many industries, including finance, have already leveraged smart contracts to introduce digital currencies. At this time, the utility industry is faced with the challenge of how to structure smart contract formation in a local energy market. Specifically, they are faced with the challenge of maintaining a balance between energy generation and demand while enabling traceability, security, and unbiased peer-to-peer energy transactions, especially within a virtual power plant. This article aims at addressing the aforementioned challenges. In particular, this article investigates how to structure the microgrids in a local energy market, and how to ensure balance and resiliency with incomplete information. Taking various generation asset dimensions and demand profiles into account, simulations are performed. A novel evolutionary computing strategy to structure the simulation is proposed. A comparison is made among random order, random selection, profit-based ranking, and evolutionary strategy for coordinating the contract formation. The discussions draw attention to each method's advantages and disadvantages in terms of their value as a strategy for forming smart contracts in a local energy market. Eesti Teadusagentuur, Grant/Award Number: PUTJD915; Ministerio de Economia y Competitividad, Grant/Award Number: TIN2017-84553-C2-2-R; Ministerio de Ciencia e Innovacion, Grant/Award Number: PID2020-113614RB-C22

A general issue in climate science is the handling of big data and running complex and computationally heavy simulations. In this paper, we explore the potential of using machine learning (ML) to spare computational time and optimize data usage. The paper analyzes the effects of changes in land cover (LC), such as deforestation or urbanization, on local climate. Along with green house gas emission, LC changes are known to be important causes of climate change. ML methods were trained to learn the relation between LC changes and temperature changes. The results showed that random forest (RF) outperformed other ML methods, and especially linear regression models representing current practice in the literature. Explainable artificial intelligence (XAI) was further used to interpret the RF method and analyze the impact of different LC changes on temperature. The results mainly agree with the climate science literature, but also reveal new and interesting findings, demonstrating that ML methods in combination with XAI can be useful in analyzing the climate effects of LC changes. All parts of the analysis pipeline are explained including data pre-processing, feature extraction, ML training, performance evaluation, and XAI. Francesco Cherubini and Xiangping Hu received support of the Norwegian Research Council through the project MitiStress (project no. 286773).

Within the scope of Industry 4.0, Automated Guided Vehicles (AGVs) are used to streamline logistics through the usage of efficient path finding methods. The current path finding methods in the industry rely on excessive usage of guidance in the shape of magnets, tapes or QR codes on the floor that the AGVs follow to reach their destinations. However, the current methods lack operational flexibility and are costly to scale in the cases of job-shop floor expansions. In this paper, a dynamic path finding method with obstacle avoidance is presented which utilizes distance measuring sensors to avoid obstacles and reach the goal destination using the most direct path as possible. Tests for functionality and multi-agent scaling have been conducted to evaluate the performance of the dynamic method in a multi-agent setting. The results show that the dynamic method scales properly and is capable of navigating multiple agents through a simulated warehouse environment autonomously and without relying on external guidance. The dynamic method is able to avoid most collisions using distance measuring sensors and multi-agent negotiation to resolve conflicts among the agents that could have resulted in potential collisions. The proposed dynamic method provides a flexible and scalable path finding method for use in Industry 4.0.

Author's accepted version (post-print). © 20XX IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. Available from 20/03/2021. This paper deals with the finite-time analysis (FTA) of learning automata (LA), which is a topic for which very little work has been reported in the literature. This is as opposed to the asymptotic steady-state analysis for which there are, probably, scores of papers. As clarified later, unarguably, the FTA of Markov chains, in general, and of LA, in particular, is far more complex than the asymptotic steady-state analysis. Such an FTA provides rigid bounds for the time required for the LA to attain to a given convergence accuracy. We concentrate on the FTA of the Discretized Pursuit Automaton (DPA), which is probably one of the fastest and most accurate reported LA. Although such an analysis was carried out many years ago, we record that the previous work is flawed. More specifically, in all brevity, the flaw lies in the wrongly ``derived'' monotonic behavior of the LA after a certain number of iterations. Rather, we claim that the property should be invoked is the submartingale property. This renders the proof to be much more involved and deep. In this paper, we rectify the flaw and reestablish the FTA based on such a submartingale phenomenon. More importantly, from the derived analysis, we are able to discover and clarify, for the first time, the underlying dilemma between the DPA's exploitation and exploration properties. We also nontrivially confirm the existence of the optimal learning rate, which yields a better comprehension of the DPA itself.

This paper proposes a novel classifier based on the theory of Learning Automata (LA), reckoned to as PolyLA. The essence of our scheme is to search for a separator in the feature space by imposing an LA-based random walk in a grid system. To each node in the grid, we attach an LA whose actions are the choices of the edges forming a separator. The walk is self-enclosing, and a new random walk is started whenever the walker returns to the starting node forming a closed classification path yielding a many-edged polygon. In our approach, the different LA attached to the different nodes search for a polygon that best encircles and separates each class. Based on the obtained polygons, we perform classification by labeling items encircled by a polygon as part of a class using a ray casting function. From a methodological perspective, PolyLA has appealing properties compared to SVM. In fact, unlike PolyLA, the SVM performance is dependent on the right choice of the kernel function (e.g., linear kernel, Gaussian kernel)—which is considered a “black art.” PolyLA, on the other hand, can find arbitrarily complex separator in the feature space. We provide sound theoretical results that prove the optimality of the scheme. Furthermore, experimental results show that our scheme is able to perfectly separate both simple and complex patterns outperforming existing classifiers, such as polynomial and linear SVM, without the need to map the problem to many dimensions or to introduce a “kernel trick.” We believe that the results are impressive, given the simplicity of PolyLA compared to other approaches such as SVM.

Spectrum cartography constructs maps of metrics such as channel gain or received signal power across a geographic area of interest using spatially distributed sensor measurements. Applications of these maps include network planning, interference coordination, power control, localization, and cognitive radios to name a few. Since existing spectrum cartography techniques require accurate estimates of the sensor locations, their performance is drastically impaired by multipath affecting the positioning pilot signals, as occurs in indoor or dense urban scenarios. To overcome such a limitation, this paper introduces a novel paradigm for spectrum cartography, where estimation of spectral maps relies on features of these positioning signals rather than on location estimates. Specific learning algorithms are built upon this approach and offer a markedly improved estimation performance than existing approaches relying on localization, as demonstrated by simulation studies in indoor scenarios., Comment: 14 pages, 12 figures, 1 table. Submitted to IEEE Transactions on Signal Processing

Implementing test suites for distributed software systems is a complex and time-consuming task due to the number of test cases that need to be considered in order to obtain high coverage. We show how a formal Coloured Petri Net model can be used to automatically generate a suite of test cases for the Paxos distributed consensus protocol. The test cases cover both normal operation of the protocol as well as failure injection. To evaluate our model-based testing approach, we have implemented the Paxos protocol in the Go programming language using the quorum abstractions provided by the Gorums framework. Our experimental evaluation shows that we obtain high code coverage for our Paxos implementation using the automatically generated test cases.

Zika virus (ZIKV), the causative agent of Zika fever in humans, is an RNA virus that belongs to the genus Flavivirus. Currently, there is no approved vaccine for clinical use to combat the ZIKV infection and contain the epidemic. Epitope-based peptide vaccines have a large untapped potential for boosting vaccination safety, cross-reactivity, and immunogenicity. Though many attempts have been made to develop vaccines for ZIKV, none of these have proved to be successful. Epitope-based peptide vaccines can act as powerful alternatives to conventional vaccines due to their low production cost, less reactogenic, and allergenic responses. For designing an effective and viable epitope-based peptide vaccine against this deadly virus, it is essential to select the antigenic T-cell epitopes since epitope-based vaccines are considered safe. The in silico machine-learning-based approach for ZIKV T-cell epitope prediction would save a lot of physical experimental time and efforts for speedy vaccine development compared to in vivo approaches. We hereby have trained a machine-learning-based computational model to predict novel ZIKV T-cell epitopes by employing physicochemical properties of amino acids. The proposed ensemble model based on a voting mechanism works by blending the predictions for each class (epitope or nonepitope) from each base classifier. Predictions obtained for each class by the individual classifier are summed up, and the class with the majority vote is predicted upon. An odd number of classifiers have been used to avoid the occurrence of ties in the voting. Experimentally determined ZIKV peptide sequences data set was collected from Immune Epitope Database and Analysis Resource (IEDB) repository. The data set consists of 3,519 sequences, of which 1,762 are epitopes and 1,757 are nonepitopes. The length of sequences ranges from 6 to 30 meter. For each sequence, we extracted 13 physicochemical features. The proposed ensemble model achieved sensitivity, specificity, Gini coefficient, AUC, precision, F-score, and accuracy of 0.976, 0.959, 0.993, 0.994, 0.989, 0.985, and 97.13%, respectively. To check the consistency of the model, we carried out five-fold cross-validation and an average accuracy of 96.072% is reported. Finally, a comparative analysis of the proposed model with existing methods has been carried out using a separate validation data set, suggesting the proposed ensemble model as a better model. The proposed ensemble model will help predict novel ZIKV vaccine candidates to save lives globally and prevent future epidemic-scale outbreaks.

Presentation at the online conference International Conference on Information-communication technologies enhanced Social Sciences and Humanities (ICTeSSH), 28.06.21 - 30.06.21, arranged by Science 2.0 Alliance. https://ictessh.pubpub.org/. In human subject research, various data about the studied individuals are collected. Through re-identification and statistical inferences, this data can be exploited for interests other than the ones the subjects initially consented to. Such exploitation must be avoided to maintain trust with the researched population. We argue that keeping data-access policies up-to-date and building accountability on research data processing can reflect subjects’ consent and mitigate data misuse. With accountability in mind, we are building Lohpi: a decentralized system for research data sharing with up-to-date access policies. We demonstrate our initial prototype with timely delivery of policy changes along with minimal access control overhead.

Abstract Contemporary energy research is becoming more interdisciplinary through the involvement of technical, economic, and social aspects that must be addressed simultaneously. Within such interdisciplinary energy research, the novel domain of energy informatics plays an important role, as it involves different disciplines addressing the socio-techno-economic challenges of sustainable energy and power systems in a holistic manner. The objective of this paper is to draw an overview of the novel domain of energy informatics by addressing the educational opportunities as well as related challenges in light of current trends and the future direction of research and industrial innovation. In this study we discuss the energy informatics domain in a way that goes beyond a purely scientific research perspective. This paper widens the analyses by including reflections on current and future didactic approaches with industrial innovation and research as a background. This paper provides key recommendations for the content of a foundational introductory energy informatics course, as well as suggestions on distinguishing features to be addressed through more specialized courses in the field. The importance of this work is based on the need for better guidelines for a more appropriate education of a new generation of experts who can take on the novel interdisciplinary challenges present in future integrated, sustainable energy systems. Article highlights Didactic approaches in the energy informatics domain are discussed based on research and industrial trends. Research trends and industrial innovation driven by energy informatics are investigated. A fundamental framework for an energy informatics course is defined together with specialized distinguishing features.

Researchers and researched populations are actively involved in participatory epidemiology. Such studies collect many details about an individual. Recent developments in statistical inferences can lead to sensitive information leaks from seemingly insensitive data about individuals. Typical safeguarding mechanisms are vetted by ethics committees; however, the attack models are constantly evolving. Newly discovered threats, change in applicable laws or an individual's perception can raise concerns that affect the study. Addressing these concerns is imperative to maintain trust with the researched population. We are implementing Lohpi: an infrastructure for building accountability in data processing for participatory epidemiology. We address the challenge of data-ownership by allowing institutions to host data on their managed servers while being part of Lohpi. We update data access policies using gossips. We present Lohpi as a novel architecture for research data processing and evaluate the dissemination, overhead, and fault-tolerance.

The developmental transition of juvenile salmon from a freshwater resident morph (parr) to a seawater (SW) migratory morph (smolt), known as smoltification, entails a reorganization of gill function to cope with the altered water environment. Recently, we used RNAseq to characterize the breadth of transcriptional change which takes place in the gill in the FW phase of smoltification. This highlighted the importance of extended exposure to short, winter-like photoperiods (SP) followed by a subsequent increase in photoperiod for completion of transcriptional reprogramming in FW and efficient growth following transfer to SW. Here, we extend this analysis to examine the consequences of this photoperiodic history-dependent reprogramming for subsequent gill responses upon exposure to SW. We use RNAseq to analyze gill samples taken from fish raised on the photoperiod regimes we used previously and then challenged by SW exposure for 24 hours. While fish held on constant light (LL) throughout were able to hypo-osmoregulate during a 24 hours SW challenge, the associated gill transcriptional response was highly distinctive from that in fish which had experienced a 7-week period of exposure to SP followed by a return to LL (SPLL) and had consequently acquired the characteristics of fully developed smolts. Fish transferred from LL to SP, and then held on SP for the remainder of the study was unable to hypo-osmoregulate, and the associated gill transcriptional response to SW exposure featured many transcripts apparently regulated by the glucocorticoid stress axis and by the osmo-sensing transcription factor NFAT5. The importance of these pathways for the gill transcriptional response to SW exposure appears to diminish as a consequence of photoperiod mediated induction of the smolt phenotype, presumably reflecting preparatory developmental changes taking place during this process.

Gastrointestinal (GI) endoscopy has been an active field of research motivated by the large number of highly lethal GI cancers. Early GI cancer precursors are often missed during the endoscopic surveillance. The high missed rate of such abnormalities during endoscopy is thus a critical bottleneck. Lack of attentiveness due to tiring procedures, and requirement of training are few contributing factors. An automatic GI disease classification system can help reduce such risks by flagging suspicious frames and lesions. GI endoscopy consists of several multi-organ surveillance, therefore, there is need to develop methods that can generalize to various endoscopic findings. In this realm, we present a comprehensive analysis of the Medico GI challenges: Medical Multimedia Task at MediaEval 2017, Medico Multimedia Task at MediaEval 2018, and BioMedia ACM MM Grand Challenge 2019. These challenges are initiative to set-up a benchmark for different computer vision methods applied to the multi-class endoscopic images and promote to build new approaches that could reliably be used in clinics. We report the performance of 21 participating teams over a period of three consecutive years and provide a detailed analysis of the methods used by the participants, highlighting the challenges and shortcomings of the current approaches and dissect their credibility for the use in clinical settings. Our analysis revealed that the participants achieved an improvement on maximum Mathew correlation coefficient (MCC) from 82.68% in 2017 to 93.98% in 2018 and 95.20% in 2019 challenges, and a significant increase in computational speed over consecutive years.

In this thesis, methods to support high level situation awareness in ship navigators through appropriate automation are investigated. Situation awareness relates to the perception of the environment (level 1), comprehension of the situation (level 2), and projection of future dynamics (level 3). Ship navigators likely conduct mental simulations of future ship traffic (level 3 projections), that facilitate proactive collision avoidance actions. Such actions may include minor speed and/or heading alterations that can prevent future close-encounter situations from arising, enhancing the overall safety of maritime operations. Currently, there is limited automation support for level 3 projections, where the most common approaches utilize linear predictions based on constant speed and course values. Such approaches, however, are not capable of predicting more complex ship behavior. Ship navigators likely facilitate such predictions by developing models for level 3 situation awareness through experience. It is, therefore, suggested in this thesis to develop methods that emulate the development of high level human situation awareness. This is facilitated by leveraging machine learning, where navigational experience is artificially represented by historical AIS data. First, methods are developed to emulate human situation awareness by developing categorization functions. In this manner, historical ship behavior is categorized to reflect distinct patterns. To facilitate this, machine learning is leveraged to generate meaningful representations of historical AIS trajectories, and discover clusters of specific behavior. Second, methods are developed to facilitate pattern matching of an observed trajectory segment to clusters of historical ship behavior. Finally, the research in this thesis presents methods to predict future ship behavior with respect to a given cluster. Such predictions are, furthermore, on a scale intended to support proactive collision avoidance actions. Two main approaches are used to facilitate these functions. The first utilizes eigendecomposition-based approaches via locally extracted AIS trajectory segments. Anomaly detection is also facilitated via this approach in support of the outlined functions. The second utilizes deep learning-based approaches applied to regionally extracted trajectories. Both approaches are found to be successful in discovering clusters of specific ship behavior in relevant data sets, classifying a trajectory segment to a given cluster or clusters, as well as predicting the future behavior. Furthermore, the local ship behavior techniques can be trained to facilitate live predictions. The deep learning-based techniques, however, require significantly more training time. These models will, therefore, need to be pre-trained. Once trained, however, the deep learning models will facilitate almost instantaneous predictions.

\textit{Diabetes} is a disease defined by raised blood glucose levels, and people with the disorder have to adhere to a strict self-management regime to avoid short- and long-term complications. Today the World Health Organization reports that over 420 million people have diabetes, and the condition is one of the most common causes of death worldwide. This thesis presents the design and implementation of a group-based motivational application for people with diabetes. The proposed application applies persuasive design techniques and aspects of motivational theory to help people stay motivated in their management of the disorder through groups. Thus, decreasing the risk for short- and long-term complications. The result is \textit{Salutem}, an application where users can define custom goals, participate in groups, and receive progression updates and notifications that encourage them to motivate each other. The thesis also shows how other applications can arise from openness with user data. For example, with other application's integration to modern technologies such as HealthKit, the proposed application gathers blood glucose data from the user's CGM through HealthKit. Because of the ongoing pandemic, an extensive user study of the implemented system is not conducted. As a result, it could not be determined if the application will affect the user's motivation. However, a user study of the application's usability is conducted on a small group of users who reported that the application had an appealing design and work as intended.

Extensive research showed that the physiological response of human tissue to exposure to low-frequency electromagnetic fields is the induction of an electric current in the body segments. As a result, each segment of the human body behaves as a relay, which retransmits the radio-frequency (RF) signal. To investigate the impact of this phenomenon on the Doppler characteristics of the received RF signal, we introduce a new three-dimensional (3D) non-stationary channel model to describe the propagation phenomenon taking place in an indoor environment. Here, the indoor space is equipped with a multiple-input multiple-output (MIMO) system. A single person is moving in the indoor space and is modelled by a cluster of synchronized moving point scatterers, which behave as relays. We derive the time-variant (TV) channel transfer function (CTF) with TV path gains and TV path delays. The expression of the TV path gains is obtained from the instantaneous total received power at the receiver side. This TV total received power is expressed as the product of the TV power of the RF signal initially transmitted and received by a body segment and the TV received power of the redirected signal. These TV powers are determined according the free-space path-loss model. Also, a closed-form approximate solution to the spectrogram of the TVCTF is derived. Here, we analyse the effect of the motion of the person and the validity of the relay assumption on the spectrogram, the TV mean Doppler shift (MDS), and the TV Doppler shift (DS) of the TVCTF. Simulation results are presented to illustrate the proposed channel model.

Security remains under-addressed in many organisations, illustrated by the number of large-scale software security breaches. Preventing breaches can begin during software development if attention is paid to security during the software’s design and implementation. One approach to security assurance during software development is to examine communications between developers as a means of studying the security concerns of the project. Prior research has investigated models for classifying project communication messages (e.g., issues or commits) as security related or not. A known problem is that these models are project-specific, limiting their use by other projects or organisations. We investigate whether we can build a generic classification model that can generalise across projects. We define a set of security keywords by extracting them from relevant security sources, dividing them into four categories: asset, attack/threat, control/mitigation, and implicit. Using different combinations of these categories and including them in the training dataset, we built a classification model and evaluated it on industrial, open-source, and research-based datasets containing over 45 different products. Our model based on harvested security keywords as a feature set shows average recall from 55 to 86%, minimum recall from 43 to 71% and maximum recall from 60 to 100%. An average f-score between 3.4 and 88%, an average g-measure of at least 66% across all the dataset, and an average AUC of ROC from 69 to 89%. In addition, models that use externally sourced features outperformed models that use project-specific features on average by a margin of 26–44% in recall, 22–50% in g-measure, 0.4–28% in f-score, and 15–19% in AUC of ROC. Further, our results outperform a state-of-the-art prediction model for security bug reports in all cases. We find using sound statistical and effect size tests that (1) using harvested security keywords as features to train a text classification model improve classification models and generalise to other projects significantly. (2) Including features in the training dataset before model construction improve classification models significantly. (3) Different security categories represent predictors for different projects. Finally, we introduce new and promising approaches to construct models that can generalise across different independent projects.

© 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. During extreme weather events like hurricanes, trees can cause significant challenges for the local communities with roadway closures or power outages. Local responders must act quickly with information regarding the extent and severity of hurricane damage to better manage recovery procedures following natural disasters. This paper proposes an approach to automatically identify fallen trees on roadways using high-resolution satellite imagery before and after a hurricane. The approach detects fallen trees on roadways via a co-voting strategy of three different algorithms and tailored dissimilarity scores. The proposed method does not rely on the large manually labeled satellite image data, making it more practical than existing approaches. Our solution has been implemented and validated on an actual roadway closure dataset from Hurricane Michael in Tallahassee, Florida, in October 2018

To date, non-wear detection algorithms commonly employ a 30, 60, or even 90 mins interval or window in which acceleration values need to be below a threshold value. A major drawback of such intervals is that they need to be long enough to prevent false positives (type I errors), while short enough to prevent false negatives (type II errors), which limits detecting both short and longer episodes of non-wear time. In this paper, we propose a novel non-wear detection algorithm that eliminates the need for an interval. Rather than inspecting acceleration within intervals, we explore acceleration right before and right after an episode of non-wear time. We trained a deep convolutional neural network that was able to infer non-wear time by detecting when the accelerometer was removed and when it was placed back on again. We evaluate our algorithm against several baseline and existing non-wear algorithms, and our algorithm achieves a perfect precision, a recall of 0.9962, and an F1 score of 0.9981, outperforming all evaluated algorithms. Although our algorithm was developed using patterns learned from a hip-worn accelerometer, we propose algorithmic steps that can easily be applied to a wrist-worn accelerometer and a retrained classification model.

Articles III and IV republished with permission of ACM (Association for Computing Machinery), from the Proceedings of the 23rd ACM/IEEE International Conference on Model Driven Engineering Languages and Systems, 2020; permission conveyed through Copyright Clearance Center,Inc. In model-driven software engineering, models are used in all phases of the development process. These models must hold a high quality since the implementation of the systems they represent relies on them. Models may get broken due to various editions throughout their life-cycle, and preserving their quality is crucial. Several existing tools reduce the burden of manually dealing with issues that affect models’ quality, such as syntax errors, model smells, and inadequate structures. However, the same issues might not have the same solutions in all contexts due to different user preferences and business policies. Personalization would enhance the usability of automatic repairs in different contexts while reusing the experience from previous repairs would avoid duplicated calculations when facing similar issues. In addition, the variety of model types together with the variety of potential issues require model repair tools and approaches which are flexible, extendible, and customizable. To this end, this thesis will focus on investigating the application of machine learning (ML) for repairing models. Our aim is to build a model repair approach that (i) provides automatic model repair, (ii) allows for user personalization, and (iii) may be extended to support the repair of different types of models which possess different kinds of issues. As a result, this thesis contains theoretical and practical contributions regarding the application of ML in model repair and the design of a personalizable and extensible model repair framework. Applying ML in model repair is not a straightforward process, as most ML algorithms require a large amount of labeled data which is still unavailable in the modeling field. Hence, we propose the use of reinforcement learning (RL) algorithms, which can learn to solve a problem by directly interacting with it, without needing to train on labeled data. To improve the performance by reusing experience from previous repairs, we also implement a transfer learning (TL) approach. In order to provide personalizable and extensible model repair, we have designed a modular framework: PARMOREL, wherewe integrate our RL and TL implementations. In PARMOREL, users can specify their repair settings before or after the repair and change the repair result by giving feedback to the framework. We extend and evaluate PARMOREL’s modules through a series of implementations. The results show that we achieve automatic and personalized model repair and that PARMOREL supports different configurations of types of models, issues, actions, preferences, and learning algorithms.

Background: Infertility affects one in six couples. New digital resources exist which enable the study of lived experience of persons with infertility. Blogging represents a forum for sharing narratives and experiences. To provide high quality care for persons with a history of infertility, it is crucial to ascertain what they value as significant in their situation. Blogs with a focus on infertility may provide this information. Objectives: The aim of this study was to gain insight into which infertility-related issues are discussed on Swedish infertility blogs. Methods: In total, 70 infertility blogs were identified on the Internet and 25 met the inclusion criteria. A quantitative–qualitative content analysis was performed with the support of the Gavagai Explorer text analysis software. Results: A total of 4508 postings were retrieved from the blogs, all of which were written by women. The outcome of the analysis resulted into the following topics: Emotions (16.8%), Relations (12.5%), Time and waiting (7.1%), Body (6.6%), Care and treatment (4.2%), Food and diet (1.4%) and Exercise (0.5%). For most topics, there was a balance between positive and negative statements, but the body topic stood out by having more negative than positive sentiment. Conclusion: By considering the topics expressed in blogs, health care personnel are provided with an opportunity to better understand the situation of individuals affected by infertility.

Articles A and B are © Springer. Reused by permission. Nowadays software has scaled to a point where it is present in every aspect of our lives. It is paramount to explore techniques that speed up the construction of software and desirably increase the reliability and quality of the final products. In particular, the adoption of software systems to support almost all kinds of tasks for the society has enforced the needs to explore the development of Domain-Specific Modelling Languages (DSMLs) to bridge the gap between clients, domain experts and software engineers. Moreover, the pervasive and concurrent nature of today's software has increased the interest in the behavioural aspects of modelling languages, facilitating the simulation, execution and verification of models prior to their actual implementation to reveal errors and potential misbehaviours. One of the most prominent fields of Model-Driven Software Engineering (MDSE) is the design and implementation of DSMLs. The creation of DSMLs involves the specification of different abstraction levels to distribute the concepts of the domain appropriately. Limitations in MDSE, especially regarding a strict number of available abstraction levels, have motivated the research and application of Multilevel Modelling (MLM). While the original motivations for MLM were mainly focused on eliminating such a strictness, in recent years the proliferation of MLM approaches has allowed to expand its practical applications in many areas, for instance, in complex and distributed systems. The complexity of these systems has promoted the specification of different modelling artefacts and sublanguages to handle different parts of the systems. This need for reusing system parts and reasoning about global properties has led to the invention of a multitude of approaches for language composition. In order to utilise MLM in such complex systems, one of the novel research paths for MLM has focused on applying language composition techniques in the MLM setting. In this thesis, we further develop MultEcore which was originally developed as an approach and a tool for the specification of multilevel modelling languages. We have also extended the underlying formalisation and the practical implementation for the execution and analysis of MLM hierarchies and the composition thereof. The execution is described using Multilevel Coupled Model Transformations (MCMTs), while the analysis is performed by utilising Maude. For the composition, we have employed an amalgamation technique for MLM hierarchies and MCMTs which has its foundation from graph transformation and category theory. In particular, MLM hierarchies are composed using multi-typing and the concept of supplementary hierarchies, while MCMTs from the component hierarchies are merged together---taking into account various merge-strategies---to multi-typed rules which are applicable on the composed MLM hierarchy. The results of the thesis are evaluated with a case study in the Coloured Petri nets (CPNs) domain.

Over the past decade, high-utility itemset mining (HUIM) has received widespread attention that can emphasize more critical information than was previously possible using frequent itemset mining (FIM). Unfortunately, HUIM is very similar to FIM since the methodology determines itemsets using a binary model based on a pre-defined minimum utility threshold. Additionally, most previous works only focused on single, small datasets in HUIM, which is not realistic to any real-world scenarios today containing big data environments. In this work, the fuzzy-set theory and a MapReduce framework are both utilized to design a novel high fuzzy utility pattern mining algorithm to resolve the above issues. Fuzzy-set theory is first involved and a new algorithm called efficient high fuzzy utility itemset mining (EFUPM) is designed to discover high fuzzy utility patterns from a single machine. Two upper-bounds are then estimated to allow early pruning of unpromising candidates in the search space. To handle the large-scale of big datasets, a Hadoop-based high fuzzy utility pattern mining (HFUPM) algorithm is then developed to discover high fuzzy utility patterns based on the Hadoop framework. Experimental results clearly show that the proposed algorithms perform strongly to mine the required high fuzzy utility patterns whether in a single machine or a large-scale environment compared to the current state-of-the-art approaches.

Health-dedicated groups on social media provide different contents and social support to their peers. Our objective is to analyze users’ engagement with health education and physical activity promotion posts according to the expressed social support and social media. All health education and physical activity promotion posts on Facebook, Twitter, and Instagram during 2017–2019 by a diabetes association were extracted. We identified the type of social support within these posts; and analysed the users’ engagement with these posts according to the type of social support and social media channel. A total of 260 posts dealing with health education (n=200) and physical activity promotion (n=60) were published. Posts promoting physical activity received 54% more likes than posts on health education (p

Industrialization, innovation, and infrastructure are important drivers in improving employment and income (United Nations, 2015b). To meet the United Nation’s Sustainable Development Goals by 2030, developing countries need to invest more in research and innovation that aims to improve their infrastructure (United Nations, 2015a). More basic infrastructure, such as roads, information and communication technology (ICT) systems, and electrical power grids, is essential to improving standards of living and protecting the environment in developing countries (ibid.). Construction researchers have found that the use of ICT increases construction efficiency. Adwan and Al-Soufi (2016) classified 21 sets of ICT technologies used in construction, including web-based platforms, Building Information Modeling (BIM) based technologies, CAD- and 3D CAD–based technologies, virtual reality, and video conferencing. Among these, BIM is an information system that facilitates digital collaboration in construction. BIM provides an environment for data sharing based on 3D models, and it has been used to increase productivity and quality of building construction. In the context of transportation infrastructure, BIM reduces errors, provides better understandings of projects, and improves project performance (SmartMarket Report, 2017). In my construction industry experience, 3D modeling was applied to infrastructure projects that include roads, bridges, and railways. 3D models of infrastructure, such as that related to railways (stations and line alignments) and bridges, was used to detect design clashes and support communication with stakeholders. More broadly, BIM implementation is rapidly increasing in the context of infrastructure construction (Bradley et al., 2016), and examples include the integration of 3D infrastructure models with other technologies, such as geographic information systems, global positioning systems, laser scanners, and photogrammetry (Bradley et al., 2016; Costin et al., 2018). Researchers have also reported on BIM applications in different phases of the infrastructure life cycles (ibid.). In general, BIM supports infrastructure construction by reducing costs and risks and increasing reliability (Costin et al., 2018). The trend of rising use shows BIM’s potential to benefit the infrastructure sector. Infrastructure is critical if developing countries are to achieve their Sustainable Development Goals, which motivated me to explore BIM implementation in that context. Since developed countries is leading in BIM use, this thesis explores whether they can learn from the experiences of more advanced countries. To identify useful lessons, I conducted a comparative analysis of BIM implementation in a developed and a developing contexts. Moreover, BIM-related innovation communities were studied to understand how BIM practices can be further strengthened in infrastructure projects. This thesis focused on the following research question: “How can the implementation of Building Information Modelling be improved to support infrastructure projects in a developing context?”