Dalarna University's logo and link to the university's website

Parkinson's disease (PD) is a neurodegenerative disorder that affects millions of people worldwide, causing various motor and non-motor symptoms. Early diagnosis of PD is crucial for timely intervention and management. Gait analysis provides insights into the motor impairments with PD, aiding in early detection. In this study, different deep learning models such as CNN, LSTM, and CNN-LSTM with varying neural network depths were explored to classify PD using gait data acquired through sensor technology. The study then compared the results of deep learning models with machine learning algorithms (Random Forest (RF) and Decision Trees (DT)). The dataset used in this study consists of 93 persons with PD and 73 healthy controls (HC) collected through sensor technology. The findings reveal that the RF algorithm achieved the highest accuracy of 96%, followed by the CNN-LSTM model of 95.49 %. © 2024 IEEE.

This study investigates the performance of Logistic Regression, k-Nearest Neighbors (KNN), and Random Forest algorithms in predicting customer churn within an e-commerce platform. The choice of the mentioned algorithms was due to the unique characteristics of the dataset and the unique perception and value provided by each algorithm. Iterative models ‘examinations, encompassing preprocessing techniques, feature engineering, and rigorous evaluations, were conducted. Logistic Regression showcased moderate predictive capabilities but lagged in accurately identifying potential churners due to its assumptions of linearity between log odds and predictors. KNN emerged as the most accurate classifier, achieving superior sensitivity and specificity (98.22% and 96.35%, respectively), outperforming other models. Random Forest, with sensitivity and specificity (91.75% and 95.83% respectively) excelled in specificity but slightly lagged in sensitivity. Feature importance analysis highlighted "Tenure" as the most impactful variable for churn prediction. Preprocessing techniques differed in performance across models, emphasizing the importance of tailored preprocessing. The study's findings underscore the significance of continuous model refinement and optimization in addressing complex business challenges like customer churn. The insights serve as a foundation for businesses to implement targeted retention strategies, mitigating customer attrition, and promote growth in e-commerce platforms.

"Moln" är ett känt ord använt för att beskriva molntjänster. Begreppet moln uppstod för längesen, och användes bland annat på 70-talet för att representera delar som ingen riktigt förstod sig på. Det är inte bara ordet moln som fortfarande orsakar förvirring, utan även dess flertaliga beskrivningar. Dessa gör det svårt att få grepp om vad molntjänster egentligen innebär. Kraven för molntjänster är flera, men man talar oftast om bred tillgänglighet, konsolidering av resurser, snabb elasticitet och mätbara, självbetjäningstjänster. Implementering av säkerhet är viktig för molntjänster för att kunna skydda dem från säkerhetshot. En vanlig men också viktig säkerhetskontroll är autentisering. Autentisering tillsammans med federationslösningar möjliggör för integration mellan molntjänster som vill dela autentisering.

Rapporten är strukturerad i olika sektioner, innehållande bakgrund till problem, syften som leder fram till frågeställningar och vald metodik. Den teoretiska biten består av en litteraturstudie om bland annat; säkerhetshot mot autentisering i molnet; skydd mot dessa; autentiseringsmetoder passande för molnet; federationslösningar. I analysen jämför rapporten skillnader och likheter mellan resultat från litteraturstudien och från flera intervjuer. I slutsatsen presenteras svar på frågeställningar och resultat från analysdelen. Integration mellan molntjänsters autentiseringslösning kan lösas av federerade identiteter, som även medför fördelarna användarvänlighet och enklare administration. Att alla identiteter samlas på en och samma plats kan dock bli ett attraktivt mål för attacker. Säkerhetshot mot molntjänster existerar i all högsta grad, dock skiljer de sig inte mycket mot traditionella tjänster. Slutsatsen påvisar att kausalitet fortfarande råder för säkerhet i molntjänster, men att förtroende har blivit ytterligare en förklaringsvariabel.

In this work, the ability to create a prototype for a throughout secure and childproof system for Android is shown. This is done by combining already available products combined with some own code. Internet traffic is being pushed through a VPN tunnel to an already existing LAN in the own home. This mitigates the risks of eavesdropping attacks and other threats, and by using DNS Filtering, unwanted web pages can be blocked beforehand.

The mobile unit can be used for calls (and with further development SMS) but incoming calls from unknown numbers, and SMS, will be blocked. With the addition of Applocker, a sort of locking mechanism for software access, the use of unwanted application will be restricted. By adding a custom App-Launcer, this restriction can be even more firm.

The complete solution brings a system where the Smartphone preferably can be handled by a child and where chosen functions can be used in their regular manner but where, predetermined, functions has been eliminated from the unit – or is only accessible by the administrator. What is needed for this prototype, besides of a Smartphone, is a working LAN in the home, and an always-on computer that acts like a server, preferably a Raspberry Pi.

To perform a computer forensic investigation can sometimes be compared to putting a puzzle together, especially if the person doing the investigation was not present at the initial stages were the IT devices was seized. To get an idea of the usage of a computer solely from the information on the hard drive can be a difficult and time consuming task. Situations that can be considered particularly complicated arise when devices have been overlooked at the raid, in other words when pieces of the puzzle are missing.During the examination of a hard drive there is often traces of removable media previously connected to the computer. These kinds of traces can be of interest during the computer forensic investigation. One aspect can be to put different devices in relation to each other, to link a digital trace to a physical action. Another possible conclusion can be that some items of relevance was missed during the raid and there for not seized.This thesis describes the goals that are relevant to strive for when conducting a computer forensic investigation. Multiple aspects was reviewed to make the result usable to a computer forensic examiner working in law enforcement.The work done was basically based on an experiment where a removable USB thumb drive was connected to a computer. The computer was subsequently examined for changes and the files identified were analysed with digital tools built for these purposes. All results were reviewed with secondary digital tools or validated with manual validation techniques.The traces from the analysis provided a method. The method is supposed to be of use when a computer forensic examiner working in law enforcement is conducting searches for removable USB storage devices that was previously connected to a computer. Finally the method was reviewed and conclusions regarding advantages and disadvantages were drawn.

Deep learning (primarily using backpropagation) and neuroevolution are the preeminent methods of optimizing artificial neural networks. However, they often create black boxes that are as hard to understand as the natural brains they seek to mimic. Previous work has identified an information-theoretic tool, referred to as R, which allows us to quantify and identify mental representations in artificial cognitive systems. The use of such measures has allowed us to make previous black boxes more transparent. Here we extend R to not only identify where complex computational systems store memory about their environment but also to differentiate between different time points in the past. We show how this extended measure can identify the location of memory related to past experiences in neural networks optimized by deep learning as well as a genetic algorithm.

Solutions to combinatorial optimization problems, such as problems of locating facilities, frequently rely on heuristics to minimize the objective function. The optimum is sought iteratively and a criterion is needed to decide when the procedure (almost) attains it. Pre-setting the number of iterations dominates in OR applications, which implies that the quality of the solution cannot be ascertained. A small, almost dormant, branch of the literature suggests using statistical principles to estimate the minimum and its bounds as a tool to decide upon stopping and evaluating the quality of the solution. In this paper we examine the functioning of statistical bounds obtained from four different estimators by using simulated annealing on p-median test problems taken from Beasley’s OR-library. We find the Weibull estimator and the 2^nd order Jackknife estimator preferable and the requirement of sample size to be about 10 being much less than the current recommendation. However, reliable statistical bounds are found to depend critically on a sample of heuristic solutions of high quality and we give a simple statistic useful for checking the quality. We end the paper with an illustration on using statistical bounds in a problem of locating some 70 distribution centers of the Swedish Post in one Swedish region. 

IEEE 802.1x is an industry standard to implement physical port level security in wired and wireless Ethernets by using RADIUS infrastructure. Administrators of corporate networks need secure network admission control for their environment in a way that adds minimum traffic overhead and does not degrade the performance of the network. This research focuses on two widely used Remote Authentication Dial In User Service (RADIUS) servers, Microsoft Network Policy Server (NPS) and FreeRADIUS to evaluate their efficiency and network overhead according to a set of pre-defined key performance indicators using Protected Extensible Authentication Protocol (PEAP) in conjunction with Microsoft Challenged Handshake Authentication Protocol version 2 (MSCHAPv2). The key performance indicators – authentication time, reconnection time and protocol overhead were evaluated in real test bed configuration. Results of the experiments explain why the performance of a particular authentications system is better than the other in the given scenario. © 2019, Zarka Private University. All rights reserved.

Penetrationstest har blivit en populär metod för organisationer att se datasäkerheten ur en angripares ögon och ofta inkluderas både system och personal av ett penetrationstest.Rapporten går igenom attackerna SQL Injection, Cross-site Scripting (XSS) och Cross-site Request Forgery (CSRF) som är specifikt för webbapplikationer och även förslag på hur man kan motverka dessa. Syftet med studien är att kontrollera om företagets webbapplikation läcker information, eller om en användare kan komma förbi autentisering eller uppgradera sina rättigheter i webbapplikationen. Vidare beskrivs även vilka verktyg som används för att inhämta information för fortsatta analyser. Diskussionen kan användas som underlag för att skapa en säkrare webbapplikation. Rapporten avslutas med en indikering om att företaget har god säkerhet, fast även ytterligare underlag för framtida undersökningar. En exekutiv sammanställning av arbetet, se Bilaga 1.

The widespread use of smartphones today, and the hardware available in these smartphones should make it possible to use these devices as a foundation for digital guides at, for example, tourist attractions with a historical connection. This report examines whether it is possible to create an Android application for smartphones that functions as a digital guide. Furthermore an attempt to delete sensitive data, in this case location data from the volatile memory on a smartphone running Android, is done. It turns out that a smartphone can be used as a foundation for a digital guide as long as the area that is to be covered by the guide is located in an area that is covered by the mobile networks and that it is possible to communicate with the GPS system. Deleting all the sensitive data from the volatile memory however, is more or less impossible.

Cybersecurity for small and medium-sized enterprises (SMEs) is becoming an increasingly important aspect in today's digital landscape. This study aims to examine and analyse requirements, recommendations, and best practices in cybersecurity for SMEs, as well as to evaluate three different firewalls as potential security solutions. The study also investigates how to effectively create simple instructions for implementing security solutions for SMEs. Through literature reviews, interviews with representatives from relevant organizations, and a technical analysis of the firewalls, the study identifies key security measures, assesses the suitability of various firewall options for SMEs, and provides suggestions for an efficient way to create short instructional videos. It is then concluded that the firewall examined in the case study that is considered best suited as a security solution is FirePower 1010. 

In recent years, knowledge graphs(KGs) have become essential tools for visualizing concepts and retrieving contextual information. However, constructing KGs for new and specialized domains like Positive Energy Districts (PEDs) presents unique challenges, particularly when dealing with unstructured texts and ambiguous concepts from academic articles. This study focuses on various strategies for constructing and inferring KGs, specifically incorporating entities related to PEDs, such as projects, technologies, organizations, and locations. We utilize visualization techniques and node embedding methods to explore the graph's structure and content and apply filtering techniques and t-SNE plots to extract subgraphs based on specific categories or keywords. One of the key contributions is using the longest path method, which allows us to uncover intricate relationships, interconnectedness between entities, critical paths, and hidden patterns within the graph, providing valuable insights into the most significant connections. Additionally, community detection techniques were employed to identify distinct communities within the graph, providing further understanding of the structural organization and clusters of interconnected nodes with shared themes. The paper also presents a detailed evaluation of a question-answering system based on the KG, where the Universal Sentence Encoder was used to convert text into dense vector representations and calculate cosine similarity to find similar sentences. We assess the system's performance through precision and recall analysis and conduct statistical comparisons of graph embeddings, with Node2Vec outperforming DeepWalk in capturing similarities and connections. For edge prediction, logistic regression, focusing on pairs of neighbours that lack a direct connection, was employed to effectively identify potential connections among nodes within the graph. Additionally, probabilistic edge predictions, threshold analysis, and the significance of individual nodes were discussed. Lastly, the advantages and limitations of using existing KGs(Wikidata and DBpedia) versus constructing new ones specifically for PEDs were investigated. It is evident that further research and data enrichment is necessary to address the scarcity of domain-specific information from existing sources.

The research presented in this thesis focuses on an innovative use of digital imaging, and the machine learning techniques to assess the body rumen fill scoring in dairy cows. This study aims to enhance the efficiency of monitoring and managing dairy cow health, which is crucial for the dairy industry's productivity and sustainability.

The primary objective was to develop an automated annotation system fore valuating rumen fill status in dairy cows using digital images extracted from recorded videos. This system leverages advanced machine learning algorithms and neural networks, aiming to mimic manual assessments by veterinarians and specialists on farms. To achieve the above objectives, this thesis made use of already existing video records from a Swedish dairy farm hosting mainly the Swedish Redand the Swedish Holstein breeds. A subset of these images were then processed, manually classified using a modified rumen fill scoring system based on visual assessment, and supervised classification algorithms were trained on 277 manually annotated images.

The thesis explored various machine learning techniques for classifying these images, including Logistic Regression, Support Vector Machine (SVM), and a Deep Neural Network using the VGG16 architecture. These models were trained, validated, and tested with a dataset that included variations in cow color patterns, aiming to determine the most effective approach for automated rumen fill scoring.The results indicated that while each model had its strengths and weaknesses, the simple logistic model was performing the best in terms of test accuracy and F1 score.

This research contributes to the field of precision livestock farming, particularly in the context of dairy farming. By automating the process of rumen fill scoring, the study aims to provide dairy farmers with a reliable, efficient, and cost-effective tool for monitoring cow health. This tool has the potential to enhance dairy cow welfare, improve milk production, and support the sustainability of dairy farming operations. However, at the current state, the model accuracy of the best model was only moderate. There is a need for further improvement of the prediction performance possibly by adding more cow images, using improved image processing, and feature engineering.

This chapter shows how elasticity testing of neural networks can greatly aid our understanding of transport systems. It examines several different pieces of work which have the common theme of using neural networks, coupled with a technique of elasticity analysis, in order to reach a better understanding of transport related problems. One of the main reasons for using neural networks is that they can easily represent complex functions, often with nonlinear interactions between different parameters. The chapter focuses on the elasticity of a single parameter with respect to a single network output. However, the elasticity technique can easily be extended to explore mutual interactions between parameters. A three-dimensional elasticity plot is shown of elasticity response against occupancy and speed. When neural networks are coupled with advanced computer visualization tools they provide an immensely powerful tool for general analysis. © V. Himanen, P. Nijkamp, A. Reggiani and J. Raitio 1998. All rights reserved.

In the past few years, Geo-spatial data quality has received increasing attention and concerns. As more and more business decisions are made based on data analytic result from geo-spatial related data, low quality data means wrong or inappropriate decisions, which could have substantial effects on a business's future. In this paper, we propose a framework that can systematically ensure and improve geo-spatial data quality throughout the whole life cycle of data.

This thesis presents a systematic review of machine learning (ML) techniques for predicting phishing attacks, a prevalent form of cybercrime exploiting human vulnerabilities through deceptive communications. With the rising frequency of internet-based fraud, developing detection systems is critical. The aim of this thesis is to explore the efficacy of various ML techniques in detecting phishing attacks. This work employs the PRISMA framework to evaluate existing research on various ML models' effectiveness in identifying phishing attempts. The review highlights significant advancements in ML methodologies that enhance phishing detection, emphasizing the strengths and limitations of models such as Random Forest, Decision Trees, and Neural Networks among others. The findings are gathered from comprehensive analyses of scholarly articles, focusing on ML’s ability to adapt to the dynamic nature of cyber threats. By comparing model performances, this study not only identifies the most effective techniques.This research underscores the necessity of continuous improvement in cyber defense technologies to keep pace with evolving cyber threats. Key contributions include a detailed comparative analysis of ML models, offering a foundation for future studies to build upon and refine cybersecurity strategies.

This work have been done in cooperation with the Swedish Armed Forces (Försvarsmakten) and presents what possibilities there are for forensic investigations of the e-book reader Amazon Kindle. In its literature study it is described how previous research in the field is very limited. The work is therefore aiming to answer what data of forensic interest a Kindle can contain and how it can be extracted, where this information is stored and if this differs between different models and firmware versions, as well as if it's enough to investigate only the part of the memory that is available for the user or if further privileges to reach the whole memory area needs to be obtained.

To do this, three different models of Kindles is filled with information. After that data images are taken off them, first on only the user partition and then on the whole memory area after a privilege escalation have been performed. Gathered data is analyzed and the result is presented.

The result shows that information of forensic interest such as notes, visited web sites and documents can be found and there is therefore a value in performing forensic investigations on Amazon Kindles. There is a difference between what information that can be found and where it's stored on the different units. The units have four partitions of which only one is accessible without privilege escalation. Because of this there is an advantage of obtaining images of the whole memory area.

In addition to the above a method for bypassing the device code of a unit and thereby getting complete access to it even though it is locked is presented.

A central challenge to evolutionary computation is enabling techniques to evolve increasingly complex target end products. Frequently, direct approaches that reward only the target end product itself are not successful because the path between the starting conditions and the target end product traverses through a complex fitness landscape, where the directly accessible intermediary states may be require deleterious or even simply neutral mutations. As such, a host of techniques have sprung up to support evolutionary computation techniques taking these paths. One technique is scaffolding where intermediary targets are used to provide a path from the starting state to the end state. While scaffolding can be successful within well-understood domains it also poses the challenge of identifying useful intermediaries. Within this paper we first identify some shortcomings of scaffolding approaches ' namely, that poorly selected intermediaries may in fact hurt the evolutionary computation's chance of producing the desired target end product. We then describe a light-weight approach to selecting intermediate scaffolding states that improve the efficacy of the evolutionary computation. © 2018 Association for Computing Machinery.

The p-median problem is often used to locate P service facilities in a geographically distributed population. Important for the performance of such a model is the distance measure.

Distance measure can vary if the accuracy of the road network varies. The rst aim in this study is to analyze how the optimal location solutions vary, using the p-median model, when the road network is alternated. It is hard to nd an exact optimal solution for p-median problems. Therefore, in this study two heuristic solutions are applied, simulating annealing and a classic heuristic. The secondary aim is to compare the optimal location solutions using dierent algorithms for large p-median problem. The investigation is conducted by the means of a case study in a rural region with an asymmetrically distributed population, Dalecarlia.

The study shows that the use of more accurate road networks gives better solutions for optimal location, regardless what algorithm that is used and regardless how many service facilities that is optimized for. It is also shown that the simulated annealing algorithm not just is much faster than the classic heuristic used here, but also in most cases gives better location solutions.

Nuclear Power Plants (NPP) undergo fault and sensitivity analysis with scenario modelling to predict catastrophic events, specifically releases of Cesium 137 (Cs-137). The purpose of this thesis is to find which of 108 input-features from Modular Accident Analysis Program (MAAP)simulation code are important, when there is large release of Cs-137 emissions. The features are tested all together and in their groupings. To find important features, the Machine learning (ML) model Random Forest (RF) has a built-in attribute which identifies important features. The results of RF model classification are corroborated with Support Vector Machines (SVM), K-Nearest Neighbor (KNN) and use k-folds cross validation to improve and validate the results, resulting in a near 90% accuracy for the three ML models. RF is successful at identifying important features related to Cs-137 emissions, by using the classification model to first identify top features, to further train the models at identifying important input-features. The discovered input-features are important both within their individual groups, but also when including all features simultaneously. The large number of features included did not disrupt RF much, but the skewed dataset with few classified extreme events caused the accuracy to be lower at near 90%.

In recent years, the market for mobile devices has increased and technological developments haveadvanced in a rapid pace. More and more communication is done through mobile and handhelddevices. Therefore, even the seizures increase by these devices, because there is often very valuableforensic information stored on them. The purpose of this work is based on a method combining aqualitative and a quantitative study examining forensic tools and techniques for handheld devices.The paper analyzed forensics description of chip-off technology and its implementation, as well asCellebrites different extraction methods are tested to see what information that is possible to extractwith each method.To seek answers, literature studies, own investigations and semi-structured interviews have beenconducted. The report is divided into different sections, where the current operating systemsAndroid and iOS, Cellebrites tools and chip-off technique are described, own investigations made,the interviews carried out and the results thereof. The results show that the type of information thatis possible to obtain, differs depending on the extraction method. Chip-off technology allows a fullphysical extraction on handheld devices that can be locked and broken, but requires some training,experience and specialized equipment.

Natural evolution keeps inventing new complex and intricate forms and behaviors. Digital evolution and genetic algorithms fail to create the same kind of complexity, not just because we still lack the computational resources to rival nature, but because (it has been argued) we have not understood in principle how to create open-ended evolving systems. Much effort has been made to define such open-endedness so as to create forms of increasing complexity indefinitely. Here, however, a simple evolving computational system that satisfies all such requirements is presented. Doing so reveals a shortcoming in the definitions for open-ended evolution. The goal to create models that rival biological complexity remains. This work suggests that our current definitions allow for even simple models to pass as open-ended, and that our definitions of complexity and diversity are more important for the quest of open-ended evolution than the fact that something runs indefinitely. © 2019 Massachusetts Institute of Technology.

Deep learning of artificial neural networks (ANNs) is creating highly functional processes that are, unfortunately, nearly as hard to interpret as their biological counterparts. Identification of functional modules in natural brains plays an important role in cognitive and neuroscience alike, and can be carried out using a wide range of technologies such as fMRI, EEG/ERP, MEG, or calcium imaging. However, we do not have such robust methods at our disposal when it comes to understanding functional modules in artificial neural networks. Ideally, understanding which parts of an artificial neural network perform what function might help us to address a number of vexing problems in ANN research, such as catastrophic forgetting and overfitting. Furthermore, revealing a network's modularity could improve our trust in them by making these black boxes more transparent. Here, we introduce a new information-theoretic concept that proves useful in understanding and analyzing a network's functional modularity: the relay information IR. The relay information measures how much information groups of neurons that participate in a particular function (modules) relay from inputs to outputs. Combined with a greedy search algorithm, relay information can be used to identify computational modules in neural networks. We also show that the functionality of modules correlates with the amount of relay information they carry.

Artificial neural networks (ANNs), while exceptionally useful for classification, are vulnerable to misdirection. Small amounts of noise can significantly affect their ability to correctly complete a task. Instead of generalizing concepts, ANNs seem to focus on surface statistical regularities in a given task. Here we compare how recurrent artificial neural networks, long short-term memory units, and Markov Brains sense and remember their environments. We show that information in Markov Brains is localized and sparsely distributed, while the other neural network substrates “smear” information about the environment across all nodes, which makes them vulnerable to noise. Copyright © ALIFE 2018.All rights reserved.

Computer games are most engaging when their difficulty is well matched to the player’s ability, thereby providing an experience in which the player is neither overwhelmed nor bored. In games where the player interacts with computer-controlled opponents, the difficulty of the game can be adjusted not only by changing the distribution of opponents or game resources, but also through modifying the skill of the opponents. Applying evolutionary algorithms to evolve the artificial intelligence that controls opponent agents is one established method for adjusting opponent difficulty. Less-evolved agents (i.e., agents subject to fewer generations of evolution) make for easier opponents, while highlyevolved agents are more challenging to overcome. In this publication we test a new approach for difficulty adjustment in games: orthogonally evolved AI, where the player receives support from collaborating agents that are co-evolved with opponent agents (where collaborators and opponents have orthogonal incentives). The advantage is that game difficulty can be adjusted more granularly by manipulating two independent axes: by having more or less adept collaborators, and by having more or less adept opponents. Furthermore, human interaction can modulate (and be informed by) the performance and behavior of collaborating agents. In this way, orthogonally evolved AI both facilitates smoother difficulty adjustment and enables new game experiences. © Springer International Publishing Switzerland 2016.

Marknaden för mobila enheter och mobilitet har haft en kraftig utveckling de

senaste åren och den tekniska utvecklingen har gått snabbt fram. Grunden till en

ökad mobilitet inom förtaget ligger i att den ofta förväntas ge positiva effekter i

ökad affärsnytta och produktivitet hos de anställda. Det är ändå viktigt att komma

ihåg att det även finns risker med en ökad mobilitet. Risker som organisationer bör

ta hänsyn till. Tester gjorda i detta arbete visar att det finns mycket information

lagrad på mobila enheter. Delar av denna information kan även vara mycket

känslig information.

Smartphones och andra mobila enheters karakteristiska egenskaper, främst att de

är små och bärbara, gör att de är extra sårbara för ett flertal risker och hot. De

största hotet mot säkerheten i samband med mobila enheter anses vara risken för

dataförlust. Denna förlust sker ofta i samband med att mobila enheter avvecklas,

stjäls eller tappas bort. Ett annat stort hot är att information stjäls av eller sprids på

grund av att skadlig programkod (malware) finns på enheten.

För att komma till rätta med dessa risker och hot så behöver man titta på hur

organisationen ska hantera situationen. Det finns på marknaden säkerhetslösningar

för att hantera många delar av säkerhetsproblematiken. Mobile Device

Management (MDM) är en funktionalitet som handlar om att skapa kontroll på

tillgångar och information samt möjliggöra en säkrare användning av mobila

enheter. Ett benchmarktest och jämförelseanalys av ett urval av marknadsledande

MDM verktyg visar på att det finns ett gott stöd för flera säkerhetshöjande

funktioner.

AirWatch, MobileIron och XenMobile är tre olika MDM verktyg som alla ger ett

bra stöd för standardiserade säkerhetshöjande funktioner för mobila enheter. Att

göra ett bra val av verktyg kräver dock ett förarbete. Detta är nödvändigt för att

skapa rätta förutsättningar för ett val som stödjer organisationens behov, krav och

strategier.

Digital guides are something that starts to appear more and more these days. Several digital

guides have already been tried with mixed results. One version was developed for

“världsarvet Falun” where a PDA was inserted into a box with speakers and a GPS and short

videos gives information about certain locations in Falun. Another guide was developed for

the IPhone in which static maps were displayed with information rendered by a server. This

thesis tries to develop a server side solution for an application designed to give users a tour of

selected so called areas of interest (aoi). The server only provides coordinates and information

about aois and the specific points within. The solution uses a server with a database to store

and provide information about these areas. The first prototype for the solution uses https to

communicate with the application and SQL PDO queries to communicate with the database.

Future work with this solution could be to upgrade the server to facilitate a larger scalability

for the solution.

Many evolutionary models that explore the emergence of cooperation rely on either individual level selection or group level selection. However, natural systems are often more complex and selection is never just on the level of the individual or group alone. Here we explore how systems of collaborating agents evolve when selection is based on a mixture of group and individual performances. It has been suggested that under such situations free riders thrive and hamper evolution significantly. Here we show that free rider effects can almost be ignored. Sharing resources even with free riders benefits the evolution of cooperators, which in the long run is more beneficial than the short term cost. Copyright © ALIFE 2018.All rights reserved.

Effective password management is crucial for safeguarding online accounts and sensitive information. This research examines the current challenges and provides alternative solutions for better password management. This study encompasses a comprehensive survey and interviews conducted with individuals across various professional backgrounds. A total of 137 online users participated in the survey, which spanned over a duration of 15 days. Additionally, four individuals were interviewed to gather more indepth data. The study aimed to understand password selection behaviors and the factors influencing them. The goal is to develop practical strategies to enhance password security and mitigate unauthorized access to sensitive information.

The purpose of the study is to provide valuable insights into the complexities of password management and contribute to the development of informed approaches for stronger password security. The study emphasizes the significance of password management and highlights the importance of educating users about the risks associated with weak passwords. The findings have implications not only for the research community but also for individuals and organizations seeking to understand user behavior and attitudes towards password systems. By gaining a deeper understanding of these aspects, it becomes possible to design more effective strategies to protect online accounts and sensitive data. 

Representations, or sensor-independent internal models of the environment, are important for any type of intelligent agent to process and act in an environment. Imbuing an artificially intelligent system with such a model of the world it functions in remains a difficult problem. However, using neuro-evolution as the means to optimize such a system allows the artificial intelligence to evolve proper models of the environment. Previous work has found an information-theoretic measure, R, which measures how much information a neural computational architecture (henceforth loosely referred to as a brain) has about its environment, and can additionally be used speed up the neuro-evolutionary process. However, it is possible that this improved evolutionary adaptation comes at a cost to the brain's ability to generalize or the brain's robustness to noise. In this paper, we show that this is not the case; to the contrary, we find an improved ability of the to evolve in noisy environments when the neuro-correlate R is used to augment evolutionary adaptation. © 2019 Copyright held by the owner/author(s). Publication rights licensed to the Association for Computing Machinery.

Natural environments are full of ambient noise; nevertheless, natural cognitive systems deal greatly with uncertainty but also have ways to suppress or ignore noise unrelated to the task at hand. For most intelligent tasks, experiences and observations have to be committed to memory and these representations of reality inform future decisions. We know that deep learned artificial neural networks (ANNs) often struggle with the formation of representations. This struggle may be due to the ANN's fully interconnected, layered architecture. This forces information to be propagated over the entire system, which is different from natural brains that instead have sparsely distributed representations. Here we show how ambient noise causes neural substrates such as recurrent ANNs and long short-term memory neural networks to evolve more representations in order to function in these noisy environments, which also greatly improves their functionality. However, these systems also tend to further smear their representations over their internal states making them more vulnerable to internal noise. We also show that Markov Brains (MBs) are mostly unaffected by ambient noise, and their representations remain sparsely distributed (i.e. not smeared). This suggests that ambient noise helps to increase the amount of representations formed in neural networks, but also requires us to find additional solutions to prevent smearing of said representations. Copyright © ALIFE 2019.All rights reserved.

Theories of decision-making often posit optimal or heuristic strategies for performing a task. In optimal strategies, information is integrated over time in order to achieve the ideal outcomes; in the heuristic case, some shortcut or simplification is applied in order to make the decision faster or easier. In this article, we use a computational framework to study the evolution of both types of decision strategies in artificial agents. The fitness of these agents is assessed based on their performance on a sequential decision task where they must accurately identify the source of as many incoming information signals as they can over a finite time span. In order to examine what decision strategies evolve as a function of task characteristics, we manipulate the quality of decision information (difficulty) and the magnitude of punishments for incorrect answers. We find that trivial (but optimal) strategies evolve when punishment magnitude is lower than the reward magnitude for correct answers, and optimal information-integrating strategies evolve when either punishment magnitude is low or information quality is high. However, the computational demands of the task become much greater as information quality decreases and punishment magnitudes increase. In these cases, heuristics are used to maintain decision accuracy in spite of the limited cognitive resources agents have available. The results suggest that heuristics are an evolved response to environments with high demands on cognitive resources, where optimal strategies are particularly difficult to achieve. © 2018 American Psychological Association.

Theories of decision making are implemented in models that predict and explain behavior in terms of latent cognitive processes. But where do these models come from, and how are they instantiated in the brain? In this article, we examine several avenues where artificial intelligence (AI) and machine learning (ML) can benefit decision theory by providing new methods for developing and testing cognitive models. First, machine learning can be used to efficiently estimate the values of latent parameters in cognitive models and assign posterior probabilities to competing models of the same observed data. Second, models of decision behavior can be embedded within artificially intelligent systems to allow them to make inferences about human counterparts (goals, abilities, cognition) in real time, equipping AI with tools to interact socially. Third, AI can be used to understand how evolutionary and learning processes give rise to the cognitive abilities that support decision making. Last, the tools of experimental psychology and decision sciences can be applied to better understand the "black boxes" of neural networks by systematically testing input-output (stimulus-response) relationships. Put together, we suggest that merging ML/AI into decision-modeling-and vice versa-is a promising path toward many long-term benefits for both fields.

This paper studied public emotion and opinion concerning the opening of new IKEA stores, specifically, how much attention are attracted, how much positive and negative emotion are aroused, what IKEA-related topics are talked due to this event. Emotion is difficult to measure in retail due to data availability and limited quantitative tools. Twitter texts, written by the public to express their opinion concerning this event, are used as a suitable data source to implement sentiment analysis. Around IKEA opening days, local people post IKEA related tweets to express their emotion and opinions on that. Such “IKEA” contained tweets are collected for opinion mining in this work. To compute sentiment polarity of tweets, lexiconbased approach is used for English tweets, and machine learning methods for Swedish tweets. The conclusion is new IKEA store are paid much attention indicated by significant increasing tweets frequency, most of them are positive emotions, and four studied cities have different topics and interests related IKEA. This paper extends knowledge of consumption emotion studies of prepurchase, provide empirical analysis of IKEA entry effect on emotion. Moreover, it develops a Swedish sentiment prediction model, elastic net method, to compute Swedish tweets’ sentiment polarity which has been rarely conducted.  

The thesis has investigated the problem of information logging and storage from a forensic standpoint; and the study has been carried out towards a company which cannot be named for confidential reasons. what information should be logged and what has to be logged according to the Swedish law, and for how long time should the logs be saved. The problem here is to find out what in general should be saved into a log file so it can be applied to as many systems as possible. The thesis has also investigated when to do a forensic investigation and what to log in Microsoft server environment (webserver, firewall and exchange server).

I intend to solve the problem by describing what needs to be saved in general and find out if there is a law that demands information to be logged. I intend to find out what needs to be saved from a forensic point of view. I am going to find out what should be logged on Microsoft server applications: webserver, firewall and exchange server.

The method used to solve the problem was with practical study and literature study. Practical study where made on a Microsoft server 2012 on webserver, firewall and exchange server where I looked at the log files and what information that were saved to them.

The report finds out what should be saved in a log file from a forensic point of view and what needs to be saved according to Swedish law. The report finds how long time the log files have to be saved. The report finds when to do a forensic investigation and what to investigate.

Market research is often conducted through conventional methods such as surveys, focus

groups and interviews. But the drawbacks of these methods are that they can be costly and timeconsuming.

This study develops a new method, based on a combination of standard techniques

like sentiment analysis and normalisation, to conduct market research in a manner that is free

and quick. The method can be used in many application-areas, but this study focuses mainly on

the veganism market to identify vegan food preferences in the form of a profile.

Several food words are identified, along with their distribution between positive and negative

sentiments in the profile. Surprisingly, non-vegan foods such as cheese, cake, milk, pizza and

chicken dominate the profile, indicating that there is a significant market for vegan-suitable

alternatives for such foods. Meanwhile, vegan-suitable foods such as coconut, potato,

blueberries, kale and tofu also make strong appearances in the profile.

Validation is performed by using the method on Volkswagen vehicle data to identify positive

and negative sentiment across five car models. Some results were found to be consistent with

sales figures and expert reviews, while others were inconsistent. The reliability of the method

is therefore questionable, so the results should be used with caution.

Representations are internalmodels of the environment that can provide guidance to a behaving agent, even in the absence of sensory information. It is not clear how representations are developed and whether they are necessary or even essential for intelligent behavior.We argue here that the ability to represent relevant features of the environment is the expected consequence of an adaptive process, give a formal definition of representation based on information theory, and quantify it with a measure R. To measure how R changes over time, we evolve two types of networks-an artificial neural network and a network of hiddenMarkov gates-to solve a categorization task using a genetic algorithm. We find that the capacity to represent increases during evolutionary adaptation and that agents form representations of their environment during their lifetime. This ability allows the agents to act on sensorial inputs in the context of their acquired representations and enables complex and context-dependent behavior. We examine which concepts (features of the environment) our networks are representing, how the representations are logically encoded in the networks, and how they form as an agent behaves to solve a task. We conclude that R should be able to quantify the representations within any cognitive system and should be predictive of an agent's long-term adaptive success. © 2013 Massachusetts Institute of Technology.

Real-world data is typically dispersed among numerous businesses or governmental agencies, making it difficult to integrate them into data privacy laws like the General Data Protection Regulation of the European Union (GDPR). Two significant obstacles to the use of machine learning models in applications are the existence of such data islands and privacy issues. In this paper, we address these issues and propose ‘HFedRF: Horizontal Federated Random Forest’, a privacy-preserving federated model which is approximately lossless. Our proposed algorithm merges d random forests computed on d different devices and returns a global random forest which is used for prediction on local devices. In our methodology, we compare IIDs (Independent and Identically Distributed) and non-IIDs variant of our algorithm HFedRF with traditional machine learning (ML) methods i.e., decision tree and random forest. Our results show that we achieve benchmark comparable results with our algorithm for IID as well as non-IID settings of federated learning. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

A challenge for the clinical management of advanced Parkinson’s disease (PD) patients is the emergence of fluctuations in motor performance, which represents a significant source of disability during activities of daily living of the patients. There is a lack of objective measurement of treatment effects for in-clinic and at-home use that can provide an overview of the treatment response. The objective of this paper was to develop a method for objective quantification of advanced PD motor symptoms related to off episodes and peak dose dyskinesia, using spiral data gathered by a touch screen telemetry device. More specifically, the aim was to objectively characterize motor symptoms (bradykinesia and dyskinesia), to help in automating the process of visual interpretation of movement anomalies in spirals as rated by movement disorder specialists. Digitized upper limb movement data of 65 advanced PD patients and 10 healthy (HE) subjects were recorded as they performed spiral drawing tasks on a touch screen device in their home environment settings. Several spatiotemporal features were extracted from the time series and used as inputs to machine learning methods. The methods were validated against ratings on animated spirals scored by four movement disorder specialists who visually assessed a set of kinematic features and the motor symptom. The ability of the method to discriminate between PD patients and HE subjects and the test-retest reliability of the computed scores were also evaluated. Computed scores correlated well with mean visual ratings of individual kinematic features. The best performing classifier (Multilayer Perceptron) classified the motor symptom (bradykinesia or dyskinesia) with an accuracy of 84% and area under the receiver operating characteristics curve of 0.86 in relation to visual classifications of the raters. In addition, the method provided high discriminating power when distinguishing between PD patients and HE subjects as well as had good test-retest reliability. This study demonstrated the potential of using digital spiral analysis for objective quantification of PD-specific and/or treatment-induced motor symptoms.

Solutions to combinatorial optimization, such as p-median problems of locating facilities, frequently rely on heuristics to minimize the objective function. The minimum is sought iteratively and a criterion is needed to decide when the procedure (almost) attains it. However, pre-setting the number of iterations dominates in OR applications, which implies that the quality of the solution cannot be ascertained. In this paper we compare the methods proposed previous literate of estimating minimum, and propose some thought of it.

According to the World Health Organization, almost all human population (99%) lives in 117 countries with over 6000 cities, where air pollutant concentration exceeds recommended thresholds. The most common, so-called criteria, air pollutants that affect human lives, are particulate matter (PM) and gas-phase (SO2, CO, NO2, O3 and others). Therefore, many countries or regions worldwide have imposed regulations or interventions to reduce these effects. Whenever an intervention occurs, air quality changes due to changes in ambient factors, such as weather characteristics and human activities. One approach for assessing the effects of interventions or events on air quality is through the use of the Weather Normalized Model (WNM). However, current deterministic models struggle to accurately capture the complex, non-linear relationship between pollutant concentrations and their emission sources. Hence, the primary objective of this thesis is to examine the power of machine learning (ML) and deep learning (DL) techniques to develop and improve WNMs. Subsequently, these enhanced WNMs are employed to assess the impact of events on air quality. Furthermore, these ML/DL-based WNMs can serve as valuable tools for conducting exploratory data analysis (EDA) to uncover the correlations between independent variables (meteorological and temporal features) and air pollutant concentrations within the models. 

It has been discovered that DL techniques demonstrated their efficiency and high performance in different fields, such as natural language processing, image processing, biology, and environment. Therefore, several appropriate DL architectures (Long Short-Term Memory - LSTM, Recurrent Neural Network - RNN, Bidirectional Recurrent Neural Network - BIRNN, Convolutional Neural Network - CNN, and Gated Recurrent Unit - GRU) were tested to develop the WNMs presented in Paper I. When comparing these DL architectures and Gradient Boosting Machine (GBM), LSTM-based methods (LSTM, BiRNN) have obtained superior results in developing WNMs. The study also showed that our WNMs (DL-based) could capture the correlations between input variables (meteorological and temporal variables) and five criteria contaminants (SO2, CO, NO2, O3 and PM2.5). This is because the SHapley Additive exPlanations (SHAP) library allowed us to discover the significant factors in DL-based WNMs. Additionally, these WNMs were used to assess the air quality changes during COVID-19 lockdown periods in Ecuador. The existing normalized models operate based on the original units of pollutants and are designed for assessing pollutant concentrations under “average” or consistent weather conditions. Predicting pollution peaks presents an even greater challenge because they often lack discernible patterns. To address this, we enhanced the Weather Normalized Models (WNMs) to boost their performance specifically during daily concentration peak conditions. In the second paper, we accomplished this by developing supervised learning techniques, including Ensemble Deep Learning methods, to distinguish between daily peak and non-peak pollutant concentrations. This approach offers flexibility in categorizing pollutant concentrations as either daily concentration peaks or non-daily concentration peaks. However, it is worth noting that this method may introduce potential bias when selecting non-peak values. In the third paper, WNMs are directly applied to daily concentration peaks to predict and analyse the correlations between meteorological, temporal features and daily concentration peaks of air pollutants.

Weather Normalized Models (WNMs) are modeling methods used for assessing air contaminants under a business-as-usual (BAU) assumption. Therefore, WNMs are used to assess the impact of many events on urban pollution. Recently, different approaches have been implemented to develop WNMs and quantify the lockdown effects of COVID-19 on air quality, including Machine Learning (ML). However, more advanced methods, such as Deep Learning (DL), have never been applied for developing WNMs. In this study, we proposed WNMs based on DL algorithms, aiming to test five DL architectures and compare their performances to a recent ML approach, namely Gradient Boosting Machine (GBM). The concentrations of five air pollutants (CO, NO₂, PM_2.5, SO₂, and O₃) are studied in the city of Quito, Ecuador. The results show that Long-Short Term Memory (LSTM) and Bidirectional Recurrent Neural Network (BiRNN) outperform the other algorithms and, consequently, are recommended as appropriate WNMs to quantify the effects of the lockdowns on air pollution. Furthermore, examining the variable importance in the LSTM and BiRNN models, we identify that the most relevant temporal and meteorological features for predicting air quality are Hours (time of day), Index (1 is the first collected data and increases by one after each instance), Julian Day (day of the year), Relative Humidity, Wind Speed, and Solar Radiation. During the full lockdown, the concentration of most pollutants has decreased drastically: −48.75%, for CO, −45.76%, for SO₂, −42.17%, for PM_2.5, and −63.98%, for NO₂. The reduction of this latter gas has induced an increase of O₃ by +26.54%.

The objective of this study is to assess the efficacy of several machine learning (ML) algorithms in identifying cross-site scripting (XSS) vulnerabilities, which are a widespread and significant cybersecurity risk. Several studies have emphasized the absence of a rich data set for model training. This research employs a comprehensive dataset from open sources, which includes 219,176 scripts evenly divided into harmful and non-harmful categories. The purpose of this study is to train and evaluate the effectiveness of various machine learning approaches. The evaluation utilizes criteria such as accuracy, F1-scores, and the confusion matrix. The algorithms analyzed are support vector machines (SVM), artificial neural networks (ANN), and recurrent neural networks (RNN). Out of all the models, the Artificial Neural Network (ANN) proved to be the most efficient, with an accuracy rate of 99% and F1-scores surpassing 0.98 in all categories. It greatly outperformed the other models.

The results indicate that combining the advantages of each model with a hybrid approach could improve detection accuracy. Integrating Support Vector Machine (SVM) with Recurrent Neural Network (RNN) and Artificial Neural Network (ANN) models can provide a dependable solution. Initially, SVM can filter data, thereby reducing the analysis time. This, in turn, improves the efficiency of RNN or ANN in detecting cross-site scripting (XSS) attacks. This approach should result in a stronger detection system for XSS vulnerabilities by combining SVM's accuracy in handling non-malicious instances with the sophisticated pattern recognition abilities of RNN and ANN. 

The national railway administrations in Scandinavia, Germany, and Austria mainly resort to manual inspections to control vegetation growth along railway embankments. Manually inspecting railways is slow and time consuming. A more worrying aspect concerns the fact that human observers are often unable to estimate the true cover of vegetation on railway embankments. Further human observers often tend to disagree with each other when more than one observer is engaged for inspection. Lack of proper techniques to identify the true cover of vegetation even result in the excess usage of herbicides; seriously harming the environment and threating the ecology. Hence work in this study has investigated aspects relevant to human variationand agreement to be able to report better inspection routines. This was studied by mainly carrying out two separate yet relevant investigations.First, thirteen observers were separately asked to estimate the vegetation cover in nine imagesacquired (in nadir view) over the railway tracks. All such estimates were compared relatively and an analysis of variance resulted in a significant difference on the observers’ cover estimates (p<0.05). Bearing in difference between the observers, a second follow-up field-study on the railway tracks was initiated and properly investigated. Two railway segments (strata) representingdifferent levels of vegetationwere carefully selected. Five sample plots (each covering an area of one-by-one meter) were randomizedfrom each stratumalong the rails from the aforementioned segments and ten images were acquired in nadir view. Further three observers (with knowledge in the railway maintenance domain) were separately asked to estimate the plant cover by visually examining theplots. Again an analysis of variance resulted in a significant difference on the observers’ cover estimates (p<0.05) confirming the result from the first investigation.The differences in observations are compared against a computer vision algorithm which detects the "true" cover of vegetation in a given image. The true cover is defined as the amount of greenish pixels in each image as detected by the computer vision algorithm. Results achieved through comparison strongly indicate that inconsistency is prevalent among the estimates reported by the observers. Hence, an automated approach reporting the use of computer vision is suggested, thus transferring the manual inspections into objective monitored inspections

The goal of this work was to determine what kind of traces of user activity is left by the most popular filemanagers available on Android. This was determined by investigating their functionality, deciding onappropriate data and actions to test them and then go through them one app at a time, saving the state of filescreated by the app in-between different actions.Investigating the files that were generated by apps showed that there is a wide variety of traces depending onwhich app it was and what features it had. Some apps focus on device information while others are moreaimed at the files themselves, some even cataloguing all of the files, their sizes and location on the device.Many of them save thumbnails of pictures the user has viewed, which are kept even when that picture itselfis deleted.Not all information used in the app features was able to be gathered since the exact location they are storedwas not determined before the end of the work. However the information that was gathered was enough tostate that there is potential value in including this kind of application during a forensics investigation.