Publikationer från Linköpings universitet
Not a member yet
58266 research outputs found
Sort by
Learning and Exploiting Subgoal Structures in Classical Planning : Towards Reliable and Transparent Intelligent Agents that Learn to Plan on Multiple Levels
No full textClassical planning aims to find a plan that is a sequence of actions allowing an intelligent agent to move from its current situation to one that satisfies the goal. Finding a plan is computationally challenging. Agents in the real world often encounter structurally similar problems with varying objects but the same predicates, actions, and related goals. Generalized planning aims to find a general plan that compactly encodes efficiently obtainable plans for each problem in an infinitely large class of structurally similar problems. Thus, we can query a general plan to efficiently obtain a plan for any problem in the class. A general plan may encode behavior on different levels of abstraction. High-level abstractions include subgoal structures that encode stepping stones towards the goal. Subgoal structures play a central role in human problem-solving, enabling reasoning at a higher level before working out the details of a plan. Learning simple, compact, meaningful, and efficient subgoal structures and their hierarchies without human intervention is an open challenge. This thesis introduces a method for learning subgoal structures with a crisp characterization; they decompose problems into subproblems of controllable polynomial complexity. We represent subgoal structures using the recently introduced policy sketches language, whose simple syntax and semantics build the theoretical foundation of our work. We extend our method to address the long-standing problem of learning hierarchical policies. Our extended method iteratively decomposes classes of problems into classes of subproblems with strictly smaller polynomial complexity, resulting in effective hierarchical decompositions. Our methods learn from small example problems using combinatorial optimization. They seek the syntactically simplest solution, enabling interpretability and allowing us often to establish their correctness for an entire problem class. When learning methods fail, it often results from limited scalability or a lack of language expressivity. We develop two methods to address these limitations. First, we develop symmetry-based abstractions to reduce redundancy in training data and improve learning efficiency. Second, we develop a method for testing the language expressivity requirement of benchmark sets using first-order logic. Moreover, we take steps toward developing a scalable planning framework that avoids an exponential preprocessing step known as grounding, which is often unnecessary in generalized planning. Our framework supports expressive language features such as conditional effects and derived predicates that cannot concisely be compiled away, enabling researchers to model and address more complex planning problems
Philosophic Dialectics in the Age of Post-philosophies
No full textMy research comprises three principal areas that are interconnected with each other: The impact of political economics on the capitalist and non-capitalist epistemologies Philosophy of performativity Art as the institute of global contemporaneity ..
Digital Tools for Geometry Teaching in Upper Secondary School for Pupils with Intellectual Disabilities : An interview study with teachers from upper secondary school for pupils with intellectual disabilities
No full textAnvändningen av digitala hjälpmedel har blivit allt vanligare och viktigare i dagens skola. Det är viktigt att lärare, som är ansvariga för det pedagogiska arbetet, har god kännedom om de digitala hjälpmedlen och deras egenskaper i undervisningen. I denna studie undersöks användningen av digitala hjälpmedel för geometriundervisning inom Anpassad gymnasieskola. Syftet är att undersöka vad lärarna på Anpassad gymnasieskola anser om digitala hjälpmedel samt hur de använder dessa hjälpmedel i geometriundervisningen. Nedan presenteras studiens frågeställningar, som är kopplade till studiens ämne. Vilka digitala hjälpmedel använder lärarna för att undervisa geometri för elever på Anpassad gymnasieskola? Vad anser lärarna om digitala hjälpmedel inom geometriundervisning för elever på Anpassad gymnasieskola? På vilket sätt anpassar lärarna de digitala hjälpmedlen för att möta elevers behov i Anpassad gymnasieskola? Litteraturen som används i bakgrunden är hämtad från Skolverket samt från Linköpings universitetsbiblioteks söktjänst UNISEARCH. För insamling av data användes kvalitativa intervjuer med fyra lärare på Anpassad gymnasieskola. Lärarna gav en övervägande positiv bild av digitala hjälpmedel, där de upplevde att dessa medel ökade engagemang, självständighet och gjorde undervisningen mer tillgänglig och individanpassad. Däremot framkom även vissa utmaningar, såsom snabba progressioner, funktionalitetsbegränsningar och bristande kunskap om digitala hjälpmedel inom geometriundervisningen. Sammanfattningsvis ansåg lärarna att digitala hjälpmedel är ett bra verktyg för elever på Anpassad gymnasieskola
Objective Tracking of Gait Patterns in Trotters
No full textThis thesis proposes an objective foundation for analysing the gait patterns oftrotting racehorses using computer vision techniques, with a particular focus onidentifying gait asymmetries. The use of a Minimum Output Sum of SquaredError (MOSSE) tracker and a stereo camera system enhances the accuracy androbustness of tracking in dynamic environments. The dataset, provided by theSwedish University of Agricultural Sciences (SLU), includes videos captured un-der varying conditions using both monocular and stereo cameras. Key investi-gations address the role of image features in improving tracking effectiveness,the advantages of stereo vision over monocular setups, and the influence of fea-ture selection, video stabilisation and frame rate on tracking performance. Find-ings indicate that integrating stereo camera data and advanced image featuressignificantly improves tracking robustness, offering a reliable and objective pathforward for detecting lameness in trotting racehorses. The tested method hasthe potential to enhance equine welfare through early diagnosis and interventionwhile advancing veterinary and computer vision applications
Artificial superlattices with abrupt interfaces by monolayer-controlled growth kinetics during magnetron sputter epitaxy, case of hexagonal CrB2/TiB2 heterostructures
No full textArtificial superlattices exhibit exceptional electronic, magnetic, optical, and mechanical properties which make them unique candidates for applications in a broad range of technologies. A common key feature of superlattices is the need for atomically abrupt interfaces. However, superlattices comprised of materials with different properties, such as melting points and diffusivities, pose large challenges for achieving high crystal quality of both constituents with abrupt interfaces. By employing ion-assisted magnetron sputter epitaxy, we present an innovative solution to this problem with utilizing a unique combination of thermal radiation and kinetic energy that enable sufficient adatom mobility for epitaxial growth of both materials. The research was implemented for the case of CrB2/TiB2 heteroepitaxial superlattices, as neutron interference mirrors, wherein the constituents’ melting points differ by 1100 K. Ion-induced intermixing was avoided by commencing growth of each TiB2 and CrB2 layer by up to 3 unit cells (uc) without ion assistance, forming a buffer to protect the interface during the ion-assisted growth of the remainder of each layer. Heteroepitaxial superlattice growth with interface widths σCrB2 ∼1 uc and σTiB2 ∼2 uc was confirmed for different modulation periods. More than 3000 uc (∼1 µm) thick superlattices with abrupt interfaces were demonstrated for neutron mirror applications.Funding Agencies|Swedish National Graduate School in Neutron Scattering (SwedNess); Swedish Foundation for Strategic Research (SSF); Swedish Government Strategic Research Area in Materials Science on Advanced Functional Materials (AFM) at Linkoping University; Center in Nano-science and technology at LiTH CeNano 2021 and 2022; Society of Vacuum Coaters Foundation (SVCF); Hans Werthen Foundation; Swedish Research Council (VR) [2019-00191]; Swedish National Infrastructure in Advanced Electron Microscopy [2021-00171, RIF21-0026]; Knut and Alice Wallenberg Foundation [KAW 2015.0043]; Swedish neutron reflectometer SuperADAM at ILL [VR 2021-00159]; [GSn15-0008]; [2009 00971]</p
Defect Engineering of Mo-based Mo2-xCTz MXenes
No full textFrom the development of bronze to the doping of silicon, the discovery and enhancement of materials have played a critical role in enabling disruptive technologies. With the increasing demand for device miniaturization and sustainability, two-dimensional (2D) nanomaterials have gained significant attention since the discovery of graphene, owing to their unique morphology and size. Among these, MXenes exhibit exceptional structural and chemical diversity, making them promising for various applications, such as energy storage and catalysis. Despite substantial progress, optimizing the properties of nanomaterials remains essential for overcoming current technological challenges. One strategy for this optimization is defect engineering, which involves the intentional creation of structural defects, such as vacancies. Currently, MXene properties can be modified by altering their composition (e.g., metal elements and/or surface terminations) and introducing ordered vacancies with fixed defect fractions. This thesis presents an alternative approach: the creation of randomly distributed vacancies and vacancy clusters with tunable defect content. Paper I introduces the concept of generating random vacancies in 2D MXenes by incorporating a sacrificial element in the parent MAX phase. Specifically, the Mo2Ga2C MAX phase was alloyed with Cr and etched in an HF solution, removing both Cr and Ga atoms. The resulting Mo2-xCTz MXenes exhibited surface defects and enhanced capacitance values. Paper II further expands this approach, demonstrating that varying the Cr content in the precursor results in MXenes with different defect fractions. The limits of Cr incorporation in the precursor phases were investigated through both simulations and experiments. Additionally, 2D MXenes with varying defect concentrations were synthesized, and electrochemical characterizations indicated that the defect concentration could be optimized for superior performance. These findings suggest that this defect engineering strategy provides a viable pathway to control defect concentrations and tune MXene properties. In principle, this approach could be extended to other MAX phases and sacrificial elements, unlocking new possibilities for the development of MXenes with novel and enhanced properties. Från utveckling av brons till dopning av kisel har upptäckter och förbättringar av material haft en avgörande roll för att möjliggöra banbrytande teknologier. Med ökade krav på t.ex. allt mindre komponenter och hållbarhet har två-dimensionella (2D) material fått allt större uppmärksamhet på grund av en unik morfologi och unika egenskaper. Bland 2D-material uppvisar MXener en stor kemisk mångfald, vilket gör dem lovande för en mängd tillämpningar, såsom energilagring och katalys. Trots betydande framsteg är det viktigt att fortsätta optimera materialegenskaper, för att lösa nuvarande tekniska utmaningar. En strategi för att trimma materialegenskaper är defektdesign, som handlar om att avsiktligt skapa strukturella defekter, såsom vakanser. Nuförtiden kan egenskaper hos så kallade MXener modifieras genom att ändra materialets sammansättning (t.ex. val av metall och/eller yttermineringar) och introducera ordnade vakanser i specifika defektkoncentrationer. Denna avhandling presenterar ett alternativt tillvägagångssätt: skapandet av slumpmässigt fördelade vakanser och vakanskluster med skräddarsydd defektkoncentration. Artikel I introducerar konceptet att generera slumpmässigt fördelade vakanser i 2D MXener genom att blanda in ett extra atomslag i den ursprungliga MAX-fasen, som sedan omvandlas till 2D. För att sammanfatta, så legerades Mo2Ga2C med Cr och etsades sedan i en HFlösning, vilket avlägsnade både Cr och Ga. Detta resulterade i Mo2-xCTz MXener, med påvisade defekter och förbättrade kapacitansvärden. I Artikel II utökades detta tillvägagångssätt ytterligare, vilket visar att olika Cr-innehåll i ursprungsmaterialet resulterar i MXener med olika defektkoncentrationer. Gränserna för Cr-innehåll i MAX-fasen undersöktes genom både simuleringar och experiment. Dessutom syntetiserades 2D MXener med varierande defektkoncentrationer, och elektrokemisk karakterisering indikerade att defektkoncentrationen kunde optimeras för att få betydligt förbättrade egenskaper. Presenterade resultat tyder på defektdesign som en möjlig väg för att kontrollera defektkoncentrationer och optimera MXene-egenskaper. I princip skulle denna strategi kunna utvidgas till andra MAX-faser och andra atomslag, vilket öppnar för nya möjligheter vad gäller utveckling av MXener med nya och förbättrade egenskaper
Border times: welfare workers navigating the temporal intersections of migration law and the education system
No full textIn migration law, power asymmetry is often hidden in seemingly neutral policies where clock and calendar time enact temporal boundaries such as dates, lengths of time and chronological age, which are used for border control. In Sweden, some 8,000 youths who received deportation orders when the government introduced harsher asylum policies in 2016 were given a 'second chance' to qualify for permanent residency under the Upper Secondary School Act in 2018. To avoid deportation, they had to complete upper-secondary education and find a job within six months of graduation. In this article, I examine how the temporal governance of borders functions when migration law intersects and clashes with different legal, institutional and individual temporal boundaries and rhythms. By interviewing welfare workers who had the power to affect whether these youths would be granted residency or be deported and by analysing the use of clock time in case-law, I show how temporal boundaries can be negotiated at these intersections, while also unmasking the unpredictability of law that dictates possibilities to negotiate time. The effects of temporal intersections should be viewed as integral to the function of law and the use of clock time as a tool for border control.Funding Agencies|Helge Ax:son Johnsons stiftelse</p
ANALYZING HOW GENDER AND INTERSECTIONALITY SHAPE THE EXPERIENCES OF DISPLACED WOMEN IN REFUGEE CAMPS IN KENYA
No full textForced displacement is one of the critical humanitarian challenges facing displaced women and children globally. In Kenya, the challenge is evident in major refugee camps such as Dadaab and Kakuma, where women experience overlapping social, economic, and protection-related challenges. Despite humanitarian efforts to combat these challenges, most existing policies and programs often treat these women as a homogenous group, hence ignoring how gender intersects with other identities to create unique challenges and privileges. Therefore, this study sought to examine how gender and intersecting identities shape the experiences of displaced women living in Kenya’s refugee camps. The study also examined the social, economic, and protection-related challenges facing these women, as well as examine the extent to which intersectional frameworks have been adopted in Kenya's humanitarian policies and programs to protect these women. To thoroughly address the research objectives, a qualitative research design was adopted. In particular, secondary data was collected from highly reputable sources, including UNHCR reports, NGO publications, policy briefs, and peer-reviewed articles. Document analysis using a thematic approach was then conducted to uncover hidden recurring themes and patterns in the data. The findings showed that gender is often treated as a static vulnerability marker, leading to one-size-fits-all interventions that ignore diverse realities among displaced women. Additionally, intersectional identities are frequently silenced in policy and humanitarian discourse, resulting in disparities in access to essential services, leadership, and empowerment opportunities. Thus, the study concluded that adopting an intersectional approach in humanitarian and policy frameworks is crucial to address the multifaceted inequalities affecting displaced women. There was also a need to integrate gender and identity-responsive strategies in existing frameworks to ensure inclusivity, equity, and genuine empowerment of displaced women living in Kenya’s refugee camps
A Neural Quality Metric for BRDF Models
No full textAccurately evaluating the quality of bidirectional reflectance distribution function (BRDF) models is essential for photo-realistic rendering. Traditional BRDF-space metrics often employ numerical error measures that fail to capture perceptual differences evident in rendered images. In this paper, we introduce the first perceptually informed neural quality metric for BRDF evaluation that operates directly in BRDF space, eliminating the need for rendering during quality assessment. Our metric is implemented as a compact multi-layer perceptron (MLP), trained on a dataset of measured BRDFs supplemented with synthetically generated data and labelled using a perceptually validated image-space metric. The network takes as input paired samples of reference and approximated BRDFs and predicts their perceptual quality in terms of just-objectionable-difference (JOD) scores. We show that our neural metric achieves significantly higher correlation with human judgments than existing BRDF-space metrics. While its performance as a loss function for BRDF fitting remains limited, the proposed metric offers a perceptually grounded alternative for evaluating BRDF models
Latent Profile Analysis of Mental Health Among Children and Young Adults With Refugee Backgrounds
No full textObjectiveChildren and young adults comprise a significant proportion of the world´s refugee population and are disproportionately negatively affected by the social determinants of health. This heterogeneous group faces high rates of poor mental health, yet research investigating within-group inequalities in mental health remains limited. We performed a latent profile analysis to explore classes of mental health based on posttraumatic stress symptoms (PTSS), general functioning, and well-being. This study aimed to improve the understanding of mental health differences, thereby providing better guidance for assessment and tailored interventions.MethodThis study involved 131 children and 127 young adults with refugee backgrounds (mean age 18.21 years, 44.6% female, 23.6% unaccompanied) recruited nationwide in Sweden (2019-2022). To examine classes and their predictors, latent profile analysis was conducted, followed by multinomial logistic regression analysis.ResultsLatent profile analysis identified four distinct classes: Good Mental Health (58.1%; low PTSS, good functioning and well-being), Severe Mental Distress (13.6%; high PTSS, low functioning and well-being), Moderate Mental Strain (12.4%; low PTSS, moderate functioning, low well-being), and Resilient (15.9%, high PTSS, good functioning, moderate well-being). Social determinants of health, such as being unaccompanied, asylum status, exposure to multiple types of violence, sexual victimization, and child maltreatment, distinguished the classes.ConclusionChildren and young adults with refugee backgrounds can be categorized into classes based on clinically relevant mental health indicators. Focusing solely on those at the highest risk for poor mental health may overlook many who are mentally healthy and those who need more targeted support. Future research should aim to replicate findings and evaluate additional predictive factors at the family and societal levels.Funding Agencies|Swedish Research Council for Health, Working Life and Welfare (FORTE) [2019-12-01, 2022-01059]; Cocozza Foundation [LIU-2022-02135]</p