In today’s complexsocio-technical systems it is not possible to foresee and prepare for allfuture events. To cope with the intricacy and coupling between people,technical systems and the dynamic environment people are required tocontinuously adapt. To design resilient systems a deepened understanding ofwhat supports and enables adaptive performance is needed. In this thesis two studiesare presented that investigate how adaptive abilities can be identified andanalysed in complex work settings across domains. The studies focus onunderstanding adaptive performance, what enables successful adaptation and how contextual factors affect the performance. The first study examines how acrisis command team adapts as they lose important functions of their teamduring a response operation.  The secondstudy presents a framework to analyse adaptive behaviour in everyday work wheresystems are working near the margins of safety. The examples that underlie theframework are based on findings from focus group discussion withrepresentatives from different organisations, including health care, nuclear,transportation and emergency services. Main contributions of this thesis includethe examination of adaptive performance and of how it can be analysed as ameans to learn about and strengthen resilience. By using contextual analysis enablersof adaptive performance and its effects the overall system are identified. Theanalysis further demonstrates that resilience is not a system property but aresult of situational circumstances and organisational structures. Theframework supports practitioners and researchers in reporting findings,structuring cases and making sense of sharp-end adaptations. The analysismethod can be used to better understand system adaptive capacities, monitoradaptive patterns and enhance current methods for safety management.

Företag står inför en utmaning att kunna anpassa sig efter snabba förändringar och att producera produkter snabbt. Därför har många företag valt att implementera agila utvecklingsprocesser, där den dominanta metoden är Scrum. I Scrum finns det tre roller varav en är produktägare. Produktägare är en komplex roll som har förändrats sedan den utvecklades. Den definieras olika i populärlitteratur och auktoritativa guider vilket har lett till att företag implementerat rollen olika. I denna studie undersöks vad produktägares arbete kännetecknas av och hur de använder sin tid. Studien använder en triangulering av strukturerad observation, intervjuer och enkät för att studera de aktiviteter och roller produktägare har. Studien bygger på ledararbete och de roller som Mintzberg identifierat. Resultatet visar att produktägares arbete är hektiskt och fragmenterat. Hur en arbetsdag ser ut varierar mellan produktägare och även mellan olika dagar. Det dagarna har gemensamt är att produktägarna deltar i möten. Resultaten tyder även på att kommunikation en viktig del i arbetet. De viktigaste rollerna som har identifierats är informationsspridare, talesperson och övervakare vilka är kopplade till informationshantering. / Companies in today's market need to be able to adapt to rapid changes and to produce products quickly. Therefore, many companies have chosen to implement agile development processes, where the dominant method is Scrum. In Scrum there are three roles, one of which is product owner. Product owner is a complex role that has changed since it was developed. It is defined differently in popular literature and authoritative guides, which has led to companies have implemented the role differently. This study examines what the product owners' work is characterized by and how they spend their time. The study uses a triangulation of structured observation, interviews, and questionnaires to study the activities and roles of product owners. The study is based on managerial work and the roles that Mintzberg has identified. The result shows that the work of the product owner is hectic and fragmented. What a working day looks like varies between product owners and between different days. What these days have in common is that the product owners attend meetings. The results also indicate that communication is an important part of the work. The key roles that have been identified are disseminators, spokesperson and monitor which are linked to information management.

Den 1 juni 2022 infördes en ny lag i Rättegångsbalken vilket öppnade upp nya möjligheter för brottsbekämpande myndigheter att kunna genomsöka information som inte är lagrad internt på en kommunikationsutrustning. Denna lag fick namnet genomsökning på distans. Syftet med detta arbete var att granska lagen med koppling till de kränkningar som görs i den personliga integriteten. Vi har även undersökt avvägningen av jurisdiktion i beslutsfattningen vid inhämtning av data som inte är lagrad i Sverige samt att ett mindre experiment har genomförts för att belysa svårigheter med den praktiska tillämpningen av lagen.

Bevisvärdering sker i flera steg under en utredning och av flera olika aktörer så som utredare, forensiker och åklagare. Inom den forensiska verksamheten är värdering av resultat en central uppgift för att säkerställa rättssäkerhet. Informationstekniska resultat kan vara svåra att förstå och öppna för tolkning, vilket ger ett starkt incitament för att följa ett vetenskapligt förhållningssätt. Syftet med studien var således att undersöka hur den it-forensiska verksamheten möter värdering av informationstekniska resultat idag och vidare hur sådan kan bidra till högre kvalitet av de resultat som verksamheten levererar samt till högre rättssäkerhet. Forskningsområdet har studerats genom en litteraturstudie och semi-strukturerade intervjuer. Samtliga intervjuade respondenter besitter sakkunnighet inom det polisiära eller rättsliga området samt har flera års erfarenhet.  Studien visar att den it-forensiska verksamheten bör analysera och värdera bevisens styrka och tillförlitlighet i större utsträckning för att uppnå kvalitativa och rättssäkra resultat. Implementeringen av standardiserade vetenskapliga metoder för analys, värdering och redovisning kan minska risken för felbarhet orsakad av mänskliga resonemang samt öka spårbarheten genom förbättrad dokumentation. Dessutom skulle det ge mer enhetliga resultat och tydligare arbetsprocess för it-forensiker.

Bitcoin scams cause billions of dollars worth of damage every year, targeting both large corporations as well as individuals. A commonly used method for these scams is spam emails. These emails all share the same intention of trying to trick people into sending Bit- coin to the address attached in the emails, which can be done using various methods like threats and social engineering. This thesis investigates Bitcoin spam emails and tries to dis- tinguish the characteristics of the successful ones. The study was conducted by collecting data on 250,000+ Bitcoin addresses from emails that have all been reported as spam to the Bitcoinabuse website. These addresses were analyzed using their number of transactions and final balance, which were extracted with a Python script using Blockchain’s public API. It was found that the successful Bitcoin spam emails only made up a tiny percentage of the entire data set. Looking at the most successful subset of spams, a few key aspects were found that distinguished them from the rest. The most successful spam emails were using blackmail techniques such as sextortion and ransomware to fool their victims. This method is believed to work so well because of the emotional response it invokes from the victims, which in many cases is enough for them to fold. In addition, luck seemed to play a rather big role for the scams to work. The emails had to reach the perfect target: a per- son that would fall for the trick, have money to send as well as the knowledge to buy and transfer Bitcoin. To increase the odds of finding these types of people, the scammers send emails in large volumes.

Knowing the shapes, sizes and positional relations between features in an image can be useful for different types of image processing.Using a Distance Transform can give us these properties as a Distance Map.There are many different variations of distance transforms that can increase accuracy or add functionality, two such transforms are the Anti-Aliased Euclidean Distance Transform and the Signed Euclidean Distance Transform.To get the benefits of both of these it is of interest to see if they can be combined and if so, how does it perform?Investigating the possibility of such a transform is the main object of this thesis. To create this combined transform a copy of the image was created and then inverted, both images are transformed and the resulting distance maps are combined into one.Signed distance maps are created for three transforms using this method. The transforms in question are, EDT, AAEDT and VAAEDT.All transforms are then evaluated using a series of images containing two randomly placed circles, the circles are created using simple Signed Distance Functions. The signed transforms work and the AAEDT performs well compared to the Signed Euclidean Distance Transform.These results were expected as a similar gap in results can be seen between the regular EDT and AAEDT.But, this transform is not perfect and there is room for improvements in the accuracy, a good start for future work. / Att känna till formen, storleken samt olika objekts position relativt till varandra i en bild kan vara användbart för olika former av bildanalys.En Distance Transform kan ge oss alla dessa egenskaper i form av en ny bild, en så kallad avståndskarta.Det finns flera olika transformer som producerar avståndskartor med olika egenskaper och precision, två exempel är Anti-Aliased Euclidean Distance Transform och Signed Euclidean Distance Transform.För att få den ökade precisionen av Anti-Aliased Euclidean Distance Transform och funktionaliteten från en dubbelriktad så är det relevant att undersöka om de går att kombinera och om det går, hur presterar den nya transformen?Att undersöka om detta är möjligt är huvudmålet med detta arbete. För att skapa denna kombinerade transform så kopieras och inverteras bilden som ska transformeras, både kopian och orginalet transformeras sedan och de resulterande avståndskartorna kombineras till ett resultat.Dubbelriktade avståndskartor skapas för tre transformer, Euclidean Distance Transform, Anti-Aliased Euclidean Distance Transform och Vector Anti-Aliased Euclidean Distance Transform.Alla tre transformer utvärderas sedan genom en serie testbilder innehållandes två slumpmässigt placerade cirklar, skapade med hjälp av vektormatematik. De dubbelriktade transformerna fungerar och resultaten är i linje med motsvarande resultat för enkelriktade transformer.Detta betyder dock inte att resultaten är perfekta, utan det finns utrymme för prestandavinster i precisionen, detta är därför en bra startriktning för framtida förbättringsarbete.

Motivated by the successes in the field of deep learning, the scientific community has been increasingly interested in neural networks that are able to reason about physics. As neural networks are universal approximators, they could in theory learn representations that are more efficient than traditional methods whenever improvements are theoretically possible. This thesis, done in collaboration with Algoryx, serves both as a review of the current research in this area and as an experimental investigation of a subset of the proposed methods. We focus on how useful these methods are as \textit{learnable simulators} of mechanical systems that are possibly constrained and multiscale. The experimental investigation considers low-dimensional problems with training data generated by either custom numerical integration or by use of the physics engine AGX Dynamics. A good learnable simulator should express some important properties such as being stable, accurate, generalizable, and fast. Importantly, a generalizable simulator must be able to represent reconfigurable environments, requiring a model known as a graph neural network (GNN). The experimental results show that black-box neural networks are limited to approximate physics in the states it has been trained on. The results also suggest that traditional message-passing GNNs have a limited ability to represent more challenging multiscale systems. This is currently the most widely used method to realize GNNs and thus raises concern as there is not much to be gained by investing time into a method with fundamental limitations.

Correct certificate revocation practices are essential to each public-key infrastructure. While there exist a number of protocols to achieve revocation in PKI systems, there has been very little work on the theory behind it: Which different types of revocation can be identified? What is the intended effect of a specific revocation type to the knowledge base of each entity? As a first step towards a methodology for the development of reliable models, we present a graph-based formalism for specification and reasoning about the distribution and revocation of public keys and certificates. The model is an abstract generalization of existing PKIs and distributed in nature; each entity can issue certificates for public keys that they have confidence in, and distribute or revoke these to and from other entities. Each entity has its own public-key base and can derive new knowledge by combining this knowledge with certificates signed with known keys. Each statement that is deduced or quoted within the system derives its support from original knowledge formed outside the system. When such original knowledge is removed, all statements that depended upon it are removed as well. Cyclic support is avoided through the use of support sets. We define different revocation reasons and show how they can be modelled as specific actions. Revocation by removal, by inactivation, and by negation are all included. By policy, negative statements are the strongest, and positive are the weakest. Collisions are avoided by removing the weaker statement and, when necessary, its support. Graph transformation rules are the chosen formalism. Rules are either interactive changes that can be applied by entities, or automatically applied deductions that keep the system sound and complete after the application of an interactive rule. We show that the proposed model is sound and complete with respect to our definition of a valid state. / <p>Report code: LIU-TEK-LIC-2006:1</p>

A common classification task of today is classifying resources that consist of words. Nondiscriminative classifiers are a popular type of classifiers for such classification. This paper presents a study that determines whether a method of using clusters of words found in training data can be utilized for improved classifier performance. The study evaluates different classifier algorithms on their ability to predict waste types of products. The findings show that classifier performance improves by using clusters, even though there are many cases when the classifiers that use clusters do not outperform the classifiers that do not use clusters.

This work aims to provide an experimental evaluation of two kleptographic backdoors. The backdoors target the key generation step in Learning with Errors-based crypto algorithms and exploit freedom in choosing the seeds for the algorithm. Based on the descriptions, four variants were implemented, and their detectability was evaluated by measuring their running time and whether the output from the backdoors passed as uniformly random bit strings. Our results show that all variants of the backdoored seed generation were significantly slower than the normal routine, the fastest variant made key generation in Kyber take 10 times as long as normal. All variants did however pass the tests on uniformity. Additionally, previous work describes a countermeasure that makes the backdoors detectable. We have contributed with a description of how the detection could be implemented, and our measurements show that the detection routine makes the key generation approximately 8% slower.