This work presents a scalable solution to automate game-testing. Traditionally, game-testing has been performed by either human players or scripted Artificial Intelligence (AI) agents. While the first produces the most reliable results, the process of organizing testing sessions is time consuming. On the other hand, scripted AI dramatically speeds up the process, however, the insights it provides are far less useful: these agents’ behaviors are highly predictable. The presented solution takes the best of both worlds: the automation of scripted AI, and the richness of human testing by framing the problem within the Deep Reinforcement Learning (DRL) paradigm. Reinforcement Learning (RL) agents are trained to adapt to any unseen level and present customizable human personality traits: such as aggressiveness, greed, fear, etc. This is achieved exploring the problem from a multi-task RL setting. Each personality trait is understood as a different task which can be linearly combined by the proposed algorithm. Furthermore, since Artificial Neural Networks (ANNs) have been used to model the agent’s policies, the solution is highly adaptable and scalable. This thesis reviews the state of the art in both automatic game-testing and RL, and proposes a solution to the above-mentioned problem. Finally, promising results are obtained evaluating the solution on two different environments: a simple environment used to quantify the quality of the designed algorithm, and a generic game environment useful to show-case its applicability. In particular, results show that the designed agent is able to perform good on game levels never seen before. In addition, the agent can display any convex combination of the trained behaviors. Furthermore, its performance is as good as if it had been specifically trained on that particular combination. / Detta arbete presenterar en skalbar lösning för att automatisera speltestning. Traditionellt har speltestning utförts av antingen mänskliga spelare eller förprogrammerade agenter. Även om det förstanämnda ger de mest tillförlitliga resultaten är processen tidskrävande. Å andra sidan påskyndar förprogrammerade agenter processen dramatiskt, men de insikter som de ger är mycket mindre användbara: dessa agenters beteenden är mycket förutsägbara. Den presenterade lösningen använder det bästa av två världar: automatiseringsmöjligheten från förprogrammerade agenter samt möjligheten att simulera djupet av mänskliga tester genom att inrama problemet inom paradigmet Djup Förstärkningsinlärning. En agent baserad på förstärkningsinlärning tränas i att anpassa sig till tidigare osedda spelmiljöer och presenterar anpassningsbara mänskliga personlighetsdrag: som aggressivitet, girighet, rädsla... Eftersom Artificiella Neurala Nätverk (ANNs) har använts för att modellera agentens policyer är lösningen potentiellt mycket anpassnings- och skalbar. Denna rapport granskar först den senaste forskningen inom både automatisk speltestning och förstärkningsinlärning. Senare presenteras en lösning för ovan nämnda problem. Slutligen evalueras lösningen i två olika miljöer med lovande resultat. Den första miljön används för att kvantifiera kvaliteten på den designade algoritmen. Den andra är en generisk spelmiljö som är användbar för att påvisa lösningens tillämplighet.

Streaming video services have been ingrained into everyday life among Swedish teens, and the media content is often considered to benefit English language learners. The present study aims to verify that elevated English language proficiency and avid consumption of online streaming media appear synchronously in upper secondary school students. This is done by gauging online streaming media habits in students enrolled in a Swedish upper secondary school using a questionnaire, and then employing Pearson correlations to investigate the strength of the relationship between this data and student scores on a provided test of receptive vocabulary. Some attention is given to the effect of subtitle language choice on the viewer, as well as giving a brief summary of extramural English. The results are that there were found to be mostly weak correlations of low significance between test scores and online streaming media-use. The discussion links the predominantly weak correlations and significance values to previous studies about frequent multitasking occurring while participants are watching audiovisual media at home. Some space is given to a suggestion on how to adapt the present methodology to upper secondary schools to enable active teachers to explore how their English learners consume audiovisual streaming media and how this may relate to language proficiency. The study concludes that more research is needed to form a more accurate view of the relationship between watching online streaming audiovisual media and improved English language proficiency, and that further investigations should be of greater magnitude and breadth in both sampling as well as what demographic data is gathered.

This thesis seeks to explore the many possibilities that lie within the incorporation of the voice alongside playing the violin. This territory is still very unknown, and the author believes that many new artistic possibilities and tools are yet to be discovered.The attempts made seek to understand the difficulties and limitations of the incorporation of the voice, while discovering its bountiful treasures. / <p>"Terje Vigen" by Ivar BøksleArrangement by Olav Eirik LølandViolin and vocal - Olav Eirik Løland</p><p></p><p>Konserten börjar efter två minuter.</p>

Auditory preemption theory suggests two competing assumptions for the attention-capturing and performance-altering properties of auditory tasks. In onset preemption, attention is immediately diverted to the auditory channel. Strategic preemption involves a decision process in which the operator maintains focus on more complex auditory messages. The limitation in this process is that the human auditory, or echoic, memory store has a limit of 2 to 5 seconds, after which the message must be processed or it decays. In contrast, multiple resource theory suggests that visual and auditory tasks may be efficiently time-shared because two different pools of cognitive resources are used. Previous research regarding these competing assumptions has been limited and equivocal. Thus, the current research focused on systematically examining the effects of complexity and timing of communication interruptions on visual detection tasks. It was hypothesized that both timing and complexity levels would impact detection performance in a multi-task environment. Study 1 evaluated the impact of complexity and timing of communications occurring before malfunctions in an ongoing visual detection task. Twenty-four participants were required to complete each of the eight timing blocks that included simple or complex communications occurring simultaneously, and at 2, 5, or 8 seconds before detection events. For simple communications, participants repeated three pre-recorded words. However, for complex communications, they generated three words beginning with the same last letter of a word prompt. Results indicated that complex communications at two seconds or less occurring before a visual detection event significantly impacted response time with a 1.3 to 1.6 second delay compared to all the other timings. Detection accuracy for complex communication tasks under the simultaneous condition was significantly degraded compared to simple communications at five seconds or more prior to the task. This resulted in a 20% decline in detection accuracy. Additionally, participants' workload ratings for complex communications were significantly higher than simple communications. Study 2 examined the timing of communications occurring at the corresponding seconds after the visual detection event. Twenty-four participants were randomly assigned to the communication complexity and timing blocks as in study 1. The results did not find significant performance effects of timing or complexity of communications on detection performance. However the workload ratings for the 2 and 5 second complex communication presentations were higher compared to the same simple communication conditions. Overall, these findings support the strategic preemption assumption for well-defined, complex communications. The onset preemption assumption for simple communications was not supported. These results also suggest that the boundaries of the multiple resource theory assumption may exist up to the limits of the echoic memory store. Figures of merit for task performance under the varying levels of timing and complexity are presented. Several theoretical and practical implications are discussed.

It is no longer any secret that people worldwide are struggling with their mental health, in terms of diagnostic disorders as well as non-diagnostic measures like perceived stress. Barriers to receiving professional mental healthcare are significant, and even in locations where the availability of such care is increasing, our infrastructures are not equipped to find people the support they need. Meanwhile, in a highly-connected digital world, many people turn to outlets like social media to express themselves and their struggles and interact with like-minded others. This setting---where human experts are overwhelmed and human patients are acutely in need---is one in which we believe artificial intelligence (AI) and natural language processing (NLP) systems have great potential to do good. At the same time, we must acknowledge the limitations of our models and strive to deploy them responsibly alongside human experts, such that their logic and mistakes are transparent. We argue that models that make and explain their predictions in ways guided by domain-specific research will be more understandable to humans, who can benefit from the models' statistical knowledge but use their own judgment to mitigate the models' mistakes. In this thesis, we leverage domain expertise in the form of psychology research to develop models for two categories of emotional tasks: identifying emotional reactions in text and explaining the causes of emotional reactions. The first half of the thesis covers our work on detecting emotional reactions, where we focus on a particular, understudied type of emotional reaction: psychological distress. We present our original dataset, Dreaddit, gathered for this problem from the social media website Reddit, as well as some baseline analysis and benchmarking that shows psychological distress detection is a challenging problem. Drawing on literature that connects particular emotions to the experience of distress, we then develop several multitask models that incorporate basic emotion detection, and quantitatively change the way our distress models make their predictions to make them more readily understandable. Then, the second half of the thesis expands our scope to consider not only the emotional reaction being experienced, but also its cause. We treat this cause identification problem first as a span extraction problem in news headlines, where we employ multitask learning (jointly with basic emotion classification) and commonsense reasoning; and then as a free-form generation task in response to a long-form Reddit post, where we leverage the capabilities of large language models (LLMs) and their distilled student models. Here, as well, multitask learning with basic emotion detection is beneficial to cause identification in both settings. Our contributions in this thesis are fourfold. First, we produce a dataset for psychological distress detection, as well as emotion-infused models that incorporate emotion detection for this task. Second, we present multitask and commonsense-infused models for joint emotion detection and emotion cause extraction, showing increased performance on both tasks. Third, we produce a dataset for the new problem of emotion-focused explanation, as well as characterization of the abilities of distilled generation models for this problem. Finally, we take an overarching approach to these problems inspired by psychology theory that incorporates expert knowledge into our models where possible, enhancing explainability and performance.

Det är intressant att veta varför användare tvångsavslutar mobila appar regelbundet eftersom flera källor pekar på att det kan sluka mer batteri än om användare inte gör detta. Representanter från Google och Apple bekräftar att regelbunden tvångsavslutning på Android- och iOS-enheter kan vara batterikrävande. Relaterat till detta är hur batteriikonen är utformad i mobila operativssystem där tidigare forskning menar att användare har svårt att med säkerhet veta hur länge deras batteri kommer räcka. Det finns även forskningsrapport från 2009 som skriver att flertalet användare i deras studie inte visste att det fanns energisparande funktioner att använda sig av i operativssystemen. Min genomförda studie utgick från en förstudie som kontrollerade varför användare tvångsavslutade appar på sina mobila enheter. Det visade sig finnas flera anledningar till att användare utför regelbunden tvångsavslutning av appar på mobila enheter. Det visade sig i min förstudie att 46% av alla som deltog i förstudien tvångsavslutade appar regelbundet. Den näst största anledningen, enligt min förstudie, till regelbunden tvångsavslutning är för att användare vill ha kontroll över vilka appar som är öppna. Detta gäller för 7% alla användare som deltog. Den största anledningen till regelbunden tvångsavslutning av mobila appar är för att användarna vill spara batteri. Dessa användare utgör 28% av alla användare som deltog i förstudien. Utifrån datan från förstudien gjordes intervjuer med smartphone-användare för att ta reda på ytterligare information om hur användare tänker kring funktionen. Flera av användarna som intervjuades tvångsavslutade appar på sin smartphone och de berättar även varför de gör detta. Av de som tvångsavslutar i intervjuerna tvångsavslutar de flesta för att de tror att det sparar batteri. Denna undersökning är gjord för att få fram data kring om användare faktiskt tvångsavslutar appar på mobila enheter och varför de gör detta. Denna rapport ska vara en del i att hjälpa formgivare att designa gränssnitt utifrån denna tillhandahållna data. / I have observed some users that force-quit apps regularly. It's interesting to know why users do this because multiple sources indicate that it probably use more battery-power than not doing it. Representatives from both Google and Apple confirm that force-quitting apps regularly on Android and iOS devices may be more battery-demanding. Relative to this is how the battery-icon is designed in the mobile operative system where a research report that users have a hard time know for certainty how much battery there is left. There is also a research from 2009 that report that users in their studie did not know that there was energysaving funktions available to use in the mobile operatingsystem. My study started with a pilot study that checked why users force-quit apps on their mobile devices. There are several reasons why users force-quit apps regularly on mobile devices. It was shown in the study that 46% of all participants in the pilot-study force-quitted apps regularly on mobile devices to save battery power. My pilot-study shows that the second biggest reason for force-quitting regularly is that users want to have control over which apps are open. That is 7% of all the participants in the pilot-study. The main reason that users force-quit apps regularly is to save battery power. This category of users constitutes 28% of all the participants in the pilot-study. Based on the data from the pilot-study, interviews with smartphone users were made to find out more about what users think about this function. Multiple participants in the interviews do force-quit apps on their smartphone. They even told us why they do it. Most of the participants in the interviews that force-quit apps do it because they think it will save battery. This research is made to get data about if users force-quit apps on mobile devices and why they do it. This research report is made to help and inform designers to design interfaces from this data provided.

The use of computer games has grown exponentially in the past few years and StarCraft II is one of the most popular e-sports today, played by millions of people worldwide. This study investigates the game’s effect on skills, as perceived by the players of StarCraft II. The study was conducted by using an empirical survey and interviews. The skills being investigated were both physical and mental in their nature. The conclusion was that the majority of players perceived that their skills regarding reaction speed, multitasking, APM, analysis and strategy were increased the most, while there was a smaller change in social skills and language skills. Additionally, people who played team games of StarCraft II did, however, feel that their teamwork, conflict management and leadership capabilities increased. / Under de senaste åren har användandet av datorspel vuxit explosionsartat. StarCraft II är ett utav dagens mest populära e-sporter och spelas av miljontals människor runt om i världen. Studien undersöker om spelare uppfattar kompetensförändringar av att spela StarCraft II. Studien utfördes med hjälp utav en enkät och intervjuer. Kompetenserna vi undersökte var av både fysiska och mentala i sin karaktär. Av resultaten kan man se att majoriteten av spelarna uppfattade en ökad kompetens inom reaktionshastighet, multitasking, APM, analytisk- och strategisk förmåga, medan man ser en mindre förändring inom social kompetens och språklig kompetens. Resultaten visar även att de som valde att spela lagspel uppfattade en ökad kompetens inom samarbete, konflikthantering och ledarskap.

La thèse s’intéresse à l’impact des subventions publiques sur les décisions de production des agriculteurs. Elle analyse d’abord le lien entre subventions et efficacité technique des exploitations, puis examine les effets des subventions découplées sur la fourniture de services écosystémiques par les agriculteurs. L'influence des subventions sur le comportement des agriculteurs est une question importante dans le contexte de réformes successives des politiques agricoles. Il existe une vaste littérature sur cette question, mais celle-là présente trois limites majeures. D’une part, la littérature sur le sujet spécifique du lien entre subventions et efficacité technique repose sur une pléthore de modèles empiriques qui traitent les subventions de façon ad hoc. De plus, cette littérature est presqu’exclusivement basée sur des modèles statiques, alors que les décisions de production sont essentiellement dynamiques.D’autre part, peu de travaux analysent le rôle des subventions découplées en considérant la nature multitâche de l’agriculture. C’est pourquoi la thèse réalise tout d’abord une méta-analyse des résultats empiriques existants sur le lien entre subventions et efficacité technique, afin de contrôler les effets qui seraient dus aux méthodes utilisées. Les résultats indiquent que la modélisation des subventions comme outputs, ou l’utilisation du ratio subventions/revenu comme proxy, pourraient générer des résultats trompeurs. Ensuite, la thèse développe et estime un modèle de frontière dynamique ainsi qu’un modèle similaire mais statique. Les résultats montrent que le cadre statique surestime l'effet (négatif) des subventions sur l'efficacité technique. Enfin la thèse développe et teste un modèle d'agence multitâche indiquant que les aides découplées peuvent inciter les agriculteurs à fournir des services écosystémiques / The thesis focuses on the impact of public subsidies on farmers’ production decisions. It first concentrates on the link between public subsidies and farm technical efficiency, and then investigates the potential effects of decoupled subsidies on farmers’ provision of ecosystem services. The influence of public subsidies on farmers’ behaviour is an important policy question in the context of the successive reforms of agricultural policies. There exists an extensive literature on this question, but this literature has three main shortcomings. First, the literature on the specific topic of the subsidy-efficiency nexus relies on a plethora of empirical models in which subsidies are often treated in an ad hoc way. Second, this literature is almost exclusively based on static modelling, while most agricultural production decisions are dynamic in nature. Finally, in the literature on the incentive effects of decoupled subsidies, the multitasking nature of farming activities has receivedlittle attention. The thesis addresses these issues, first, by undertaking a meta-analysis of the existing empirical findings on the subsidy-efficiency nexus, in order to control for effects arising from the various methodologies used. The results show that modelling subsidies as outputs, or using the ratio of subsidies to farm revenue as a subsidy proxy, could lead to misleading results. Then, the thesis develops and estimates a dynamic frontier model; for comparison purposes, the static counterpart of this model is also estimated. The results indicate that the static framework overestimates the (detrimental) effect of subsidies on farm technical efficiency. Finally, the thesis develops and tests a multitasking agency model, indicating that decoupled subsidies could raise farmer's incentives to provide environmental services and ecologically sound production

Female athletes exhibit three- to six-fold greater incidence of noncontact anterior cruciate ligament (ACL) injury relative to their male counterparts. The increased risk appears to stem from interactions between several risk factors, that can roughly be categorized as anatomic, biomechanical, hormonal, and neuromuscular. Neuromuscular risk factors have recently gained a greater focus, and include differences in the timing and magnitude of activation of lower extremity and trunk musculature. In addition to neuromuscular risk factors, the incidence of ACL injury is not evenly distributed across the menstrual cycle, suggesting that hormonal fluctuations may influence neuromuscular control. Finally, it is known that even well-trained athletes experience decrements in performance and postural control when forced to attend to multiple sensory stimuli, which is common in many sports. PURPOSE: To explore neuromuscular differences in the ways healthy and ACL-injured women respond to a secondary task requiring fine motor control and sustained mental focus (typing task). Our investigation encompassed three broad aims. First, we sought to determine whether ACL-injured individuals demonstrated similar reflex profiles to healthy individuals, as well as to determine whether the ACL-involved limb was similar to its uninvolved counterpart. Our second aim was to determine whether the typing task resulted in attenuated Hoffmann (H) reflex amplitudes, and to investigate whether any observed changes were similar in healthy and ACL-injured groups. Finally, our third broad aim was to utilize more complex H reflex analysis techniques to determine whether differences in spinal excitability existed at different points in the menstrual cycle. METHODS: Thirty nine recreationally active women (20 with prior unilateral noncontact ACL injury: 24.0 ± 4.5 years; 23.8 ± 4.5 kg•m⁻²; 4.1 ± 2.6 years post-injury; 19 with no history of knee injury: 23.8 ± 4.5 years; 23.1 ± 2.3 kg•m⁻²) agreed to participate, and were tested during days 2-5 (follicular phase) of the menstrual cycle. A sub-set of this original group (n=8; 24.0 ± 4.8 years; 22.0 ± 2.1 kg•m⁻²) also agreed to return for a second testing session 24-96 hours after ovulation (early luteal phase), in order to assess H reflex differences across the menstrual cycle. During each testing session, H reflex testing was used to explore spinal-level control mechanisms of the lower extremity musculature under both Rest and Task conditions. In the control group, the dominant limb was tested (CON-D) while in the ACL group, both the uninvolved (ACL-UN) and involved (ACL-INV) limbs were assessed. Differences between groups (Control vs. ACL) and within-groups (ACL-UN vs. ACL-INV) were explored. RESULTS: At rest, H reflex parameters in ACL-INV were generally similar to ACL-UN and to CON-D. However, differences in presynaptic inhibition were apparent in ACL-INV that imply reduced reflex plasticity. During the typing task, both the Control and ACL groups experienced attenuated H reflex parameters. In the sub-set of participants who were tested twice during the menstrual cycle, a significant increase in presynaptic inhibition was observed during the early luteal phase compared to the follicular phase. CONCLUSION: While individuals with prior ACL injury display similar H reflex profiles to healthy individuals, the ACL-involved limb may demonstrate less reflex plasticity in response to environmental changes. This lack of plasticity may potentially increase the risk of re-injury. In addition, an upper extremity task requiring fine motor control and sustained mental focus attenuates the H reflex in both groups. This attenuation has implications for lower-extremity neuromuscular control in dual-task environments. Finally, the increase in presynaptic inhibition observed during the early luteal phase may provide insight into why ACL injuries are not evenly distributed across the menstrual cycle. / Graduation date: 2012

Assembly processes are an integral part of many industries, including manufacturing and production. These processes typically involve the use of robots and automated equipment to perform tasks such as picking, placing, and joining components. One solution is Mixed Reality (MR), which combines virtual and real-world elements to create an immersive environment for the operator. MR technology can be used to guide operators through the assembly process, providing real-time feedback and instructions, as well as allowing them to program the assembly process and adjust as needed.  The project was focused on developing a user interface for the Hololens 2 glasses that would allow operators to select different tools and robots and configure targets and processes for an assembly station. The team also developed a system to send information about targets, paths, and joint values to the virtual and real robot, which allowed operators to easily program the robot to perform the assembly process. It was possible to develop and test the MR system in a real-world assembly setting, evaluating its effectiveness in improving the efficiency and accuracy of the process. This project wants to demonstrate the potential of MR technology for improving assembly processes and to provide a proof-of-concept for future development in this field. / <p>Utbytesstudenter</p>