Scheduling lectures, exams, seminars etc. for a university turns out to be a harder task than what it seems to be at first glance. This problem is known as the University Course Timetabling Problem (UCTP). The UCTP has been hosted for a number of competitions throughout the years by an organization called Practice and Theory of Automated Timetabling (PATAT). Because of these competitions, the problem has been given a standard description and set of constraints as well as standard problem instances for easier comparison of research and work on the subject. However, setting a standard like this have a major drawback; no variety is introduced since new research for finding the greatest method to solve the UCTP is forced to focus on a specific set of constraints, and algorithms developed will only be optimized with these constraints in consideration. In this research we compared five well known UCTP algorithms with the standard set of constraints to a different set of constraints. The comparisons showed a difference in the rank of performance between the algorithms when some constraints were changed to fit a certain need. The differences were not great but big enough to state that previous research declaring what algorithms are best for the UCTP problem cannot be relied upon unless you use close to identical sets of constraints. If the goal is to find the best algorithm for a new set of constraints then one should not rely on a single previously defined great algorithm but instead take two or three of the top performing ones for the greatest chance of finding the most optimized solution possible. / Schemaläggning av föreläsningar, tentamen, seminarier etc. för ett universitet visar sig vara en svårare uppgift än vad det verkar vid första anblicken. Detta problem är känt som University Course Timetabling Problem (UCTP). UCTP har varit centralt i ett antal tävlingar genom åren av organisationen Practice and Theory of Automated Timetabling (PATAT). På grund av dessa tävlingar har problemet fått en standardbeskrivning och en uppsättning specifika begränsningar samt standard problemdata för enklare jämförelse av forskning och arbete i ämnet. Att sätta denna typ av standard har dock en stor nackdel; ingen variation tillförs då ny forskning för att hitta den bästa optimeringsmetoden inom UCTP tvingas att fokusera på en specifik uppsättning begränsningar och algoritmer som utvecklas kommer då endast att optimeras med dessa begränsningar i beaktande. I den här rapporten jämförde vi fem välkända UCTP algoritmer med standarduppsättningen av begränsningar mot en annan uppsättning begränsningar. Jämförelserna visade en skillnad i prestationsordningen mellan algoritmerna när vissa begränsningar ändrats för att passa ett visst behov. Skillnaderna var inte enorma men tillräckligt stora för att påvisa att tidigare forskning som förklarar vilka algoritmer som är bäst för UCTP-problemet ej är pålitlig om du inte använder nära till identiska uppsättningar av begränsningar. Om målet är att hitta den bästa algoritmen för en ny uppsättning begränsningar, bör man inte lita på en tidigare definierad effektiv algoritm utan istället använda sig utav två eller tre av de starkaste algoritmerna för den största chansen att hitta den mest optimerade lösningen.

Crowd Simulators are used to see how large virtual crowds behave. They are mostly used to simulate crowd behaviors. One of the constraints with crowd simulators is the level of detail that should be used to still have a realistic simulation. This thesis explores the idea of using imposters as a method of lowering the level of detail. Imposters are 3D-models in a scene that are only rendered as two-dimensional objects when they are far enough from the camera. The general problem statement of this thesis is to see how well people can differentiate between 3D-models and imposters. This is tested with the underlying problem how presentation affects how one interpret imposters. Presentation is tested in regards to the distance the camera have from the imposter and from what angle you look at imposters from. Using Unity 3D, an implementation of imposters was created that can capture an imposter, render an imposter and make the imposter move in different camera angles. A user study was conducted to test how well this implementation works in regard to presentation. The study consisted of several movie clips showing an imposter and its original model walking down a road. Users had to push a button when they saw through the imposter. From the user study, it was shown that distance from the camera do affect how one sees an imposter. Almost all of the other factors does not have a significant difference from each other. Users attest to that they were specifically looking for any visual artifacts and aliasing in the imposter to find any faults in them. It would have been interesting to try this implementation in a real crowd simulator. It would help to enable using other camera angles as factors to have more extreme cases to compare to. The recommendation is to try to not use imposters outside of crowd simulators, as it is too easy to see a difference between imposters and their models if the imposter is too close. / Simulatorer av folkmassor används för att se hur stora virtuella folkmassor beter sig. De används främst för att testa hur folk beter sig i verkligheten. En av begränsningarna med denna typ av simulatorer är att kunna välja hur mycket detaljer man vill begränsa i scenen. Simulationen måste gå fort men samtidigt bibehålla realismen. Detta arbete har utforskat imposters som metod för att sänka detaljer i scener. Imposters är 3D-modeller som renderas som tvådimensionella objekt när de är tillräckligt långt bort ifrån kameran. Det generella problemet som presenteras i denna rapport är att se hur väl man kan se skillnad mellan 3d-modeller och imposters. Detta testas med det underliggande problemet, hur det påverkar presentationen av imposter hur man tolkar dessa. Tolkningsfaktorer är hur långt ifrån kameran är från imposters samt vilken vinkel man ser imposters från. Studien gick till så att användare tittade på ett antal klipp som visar en imposter och dess normala modell som går på en väg. Användare skulle trycka på en knapp när de trodde de såg igenom impostern. Resultatet från studien visar att avståndet från kameran påverkar hur man synar en imposter. Nästan alla andra faktorer har ingen signifikant skillnad mellan varandra. Användare menar att de tittade specifikt efter visuella bieffekter (så som aliasing och visual artifacts) för att se några fel i imposters. Det hade varit intressant att testa denna implementation i en äkta simulator för folkmassor. Detta hade kunnat göra så flera andra kameravinklar hade kunnat användas som faktorer, något som hade skapat mer extrema fall att jämföra med. Rapportens rekommendation är att inte använda imposters utanför simulatorer av folkmassor, då det är för enkelt att se skillnaden på en imposter och dess modell om impostern är för nära.

When interacting with fictional environments, the users' sense of immersion can be broken when characters act in mechanical and predictable ways. The vast majority of AIs for such fictional characters, that control their actions, are statically scripted, and expert players can learn strategies that take advantage of this to easily win challenges that were intended to be hard. Games can also be too hard or too easy for certain players. Through the means of Reinforcement Learning, we propose a method to train adversaries in a simple environment for a game of tag from the PettingZoo library, exploring the possibility of such modern AIs to learn during the game. Our work aims towards a new concept of continuously learning AIs in video games, giving a framework to greatly increase adaptability of products to their users, and replayability of the challenges offered in them. We found that our solution allows the agents to learn during the game, but that more work should be done to achieve a model that tailors the behavior to the specific player. Nonetheless, this is an exploratory step towards more research on this new concept, which could have numerous applications in many genres of video games.

The study of multivariate networks (MVNs, i.e., large data sets where datapoints have relations to other data points and both these relations and the pointsthemselves can have attributed data) is an important task in many different fields,such as social networks for the humanities, citation networks for bibliometricsand biochemical networks for life sciences. Furthermore, when dealing withvisualization and analysis of MVNs, many open challenges still exist regardingboth computational aspects (i.e., the challenge of computing different metricsof a large-scale MVN) and visual aspects (i.e. the challenge of displaying allthe information of a large-scale MVN in a way that is comprehensible to theuser). In the search for efficient and scalable visual analytics methods, especiallyfor exploratory data analysis, this thesis explores a novel approach of aspectdrivenMVN embedding and the use of ensembles of embeddings for multi-levelsimilarity calculations. Starting from the observation that there already existseveral different embedding techniques for datatypes that are common for realworldMVNs, the main question that we will try to answer is: “Could the useof multiple embeddings provide for new and better solutions for visual analytics onmultivariate networks?" This main question then inspires the formulation of fourmore specific research goals regarding: (1) methods for combining embeddings,(2) the development of a general methodology framework, (3) new visualizationmethods, and (4) proof-of-concept applications for real-world scenarios.The focus of our work lies on similarity-based analysis within the domainsof bibliometrics and scientometrics, and our first major step is to developa methodology for combining several different embeddings (for the sameunderlying data) to augment the quality of similarity calculations. This stepincludes an adaptation of some of the key ideas from ensemble methods to thefield of embeddings, and also an interactive optimization process for finding thebest performing ensembles. Upon this foundation, we develop an aspect-drivenapproach which seeks to divide an underlying MVN into separately embeddableaspects, which in turn allows for the resulting embedding vectors to be used inflexible analysis scenarios with high level of interaction. We then proceed toshow how the concept of similarity-based analysis can be used to obtain valuableinsights to, and a better understanding of, a large set of scientific publications.For this, we introduce the abstract concept of similarity patterns which we use toexpress how a specific set of similarity criteria are distributed over a data set.Furthermore, we present proof-of-concept applications which are designed toallow the user to exploit these similarity patterns at different levels of detail. Wealso show that our proposed methodology is generalizable beyond the scope ofMVNs, and therefore could be applied to other fields as well.

LoRaWAN är ett 4G liknande nätverk som används för en billig och energisnålnätverkslösning för Internet of Things produkter. I Skellefteå består det nuvarande nätverk avtre stycket Kerlink iBTS gateways som utgör hela Skellefteås LoRaWAN nät.Detta nätverk kräver omtanke och behöver uppdateras, då det i nuläget ej fungerar felfritt.Därför har jag under projektets gån uppdaterad felsökt och lagt fram uppdateringar pånätverket för dets framtid. Projektet innefattar att en gateway har uppdaterats men i sambandmed detta slutat att fungera. Problemet uppstod av en fel implementation med statiskaadresser på samtliga gateways istället för DHCP adresser som enligt protokoll ska användas.Nyaste uppdateringen stödjer därför inte längre statiska adresser och därför byttes den tillDHCP och slutade att fungera. Efter genomfört projekt fungerar nu LoRaWAN nätet bättresamt samtliga gateways är uppdaterade.

No description available.

No description available.

<p>QC 20130530</p>

No description available.

The measurement of water-levels is critical within hydropower production and with already existing camera surveillance in abundance for the purpose of manual supervision. The allure of automatic visual reading to replace the need for manual oversight is significant in the pursuit of fully data driven solutions within hydropower systems. Could images of water level scales along with machine learning functionality produce a reliable and feasible solution? There are many aspects of visually reading any water-level in practice, such as lighting conditions, environmental interference. Great water level fluctuation needs to be overcome by providing an expansive and diverse dataset based on high resolution image capture. The provided solution is based on machine learning algorithms such as two- dimensional convolution, computationally performed and trained by a high power desktop computer. This algorithm is deployed in the field on a low power System-on-a-Chip (SoC) computer with dedicated in system high resolution camera. Basic image manipulation is performed in algorithm to eliminate image noise and to focus on level scale region of interest. The provided solution overcomes the issues at hand and results in a tested proof of concept system capable of ±5mm level reading accuracy with reliability of up to ≥ 99%, within a predefined data range. The results prove that the solution is feasible and a system implementing it or a derivative solution is practically implementable for real life use cases at edge locations.