Examensarbeten för masterexamen


Senast publicerade

Visar 1 - 5 av 395
  • Post
    Incorporating Interior Property Images for Predicting Housing Values
    (2024) Gortzak, Adrian; Ulusoy, Nedim Can; Chalmers tekniska högskola / Institutionen för matematiska vetenskaper; Särkkä, Aila; Malekipirbazari, Milad
    The property valuation process for the real estate market is essential for predicting a fair market value. This process is traditionally carried out by brokers, including inspecting and assessing the subject property to find comparable sales for comparative market analysis (CMA). Meanwhile, an automated valuation model (AVM) can help achieve an autonomous version of this process, which speeds up the process but lacks some of the inputs that a manual assessment provides. AVMs have difficulty considering more subjective architectural qualities, such as beauty, stability, and utility, due to the difficulty of quantifying these aspects objectively. New advancements in Visual Transformers (ViT), self-supervised learning and Contrastive Language- Image Pre-training (CLIP) technologies have shown favourable improvements in the field of computer vision. Therefore, this study explores the potential improvements of these new techniques within the visual feature extraction task to enhance the AVMs from interior images. By applying ViTs as binary classifiers, clusters, and textual descriptions matching, we aim to enrich the feature extraction process for a property valuation model in the region of Uppsala County, Sweden. Our findings show modest enhancements in the AVM’s performance, which align with prior studies, but also highlight that these new technologies can extract more detailed features compared to previous methods. Furthermore, they demonstrate the potential for these technologies to capture more comprehensible architectural qualities from images, which could significantly assist brokers in the valuation process.
  • Post
    Hypergeometric Functions and Their Generalizations to Higher Dimensions: A study of the classical hypergeometric function and its generalizations associated with root systems
    (2024) Johansson, Oskar; Chalmers tekniska högskola / Institutionen för matematiska vetenskaper; Hallnäs, Martin; Hallnäs, Martin
    The hypergeometric differential equation is a classical ODE of second order, and it was already studied by Gauss. The hypergeometric function is classically defined as the solution to this equation that is analytic at x = 0. With this definition it is not obvious how to generalize the hypergeometric function to higher dimensions. With a shift in perspective we can arrive at the same differential equation by studying a certain eigenvalue problem of polynomials of so called Dunkl operators. These are easier to generalize and will lead us to hypergeometric functions associated with so called root systems in higher dimension.
  • Post
    Studying Genetic Diversity and Evolutionary Pattern in Human Immunodeficiency Virus: Utilizing Sequencing Data and Machine learning
    (2024) Varghaei, Laleh; Chalmers tekniska högskola / Institutionen för matematiska vetenskaper; Kristiansson, Erik; Lorén, Erik
    The Acquired Immunodeficiency Syndrome (AIDS) pandemic has affected millions of people worldwide and posed a threat to global health. Since the discovery of the Human Immunodeficiency Virus (HIV) as the cause of the AIDS pandemic, numerous studies have been conducted on this virus, and many attempts have been made to develop an effective treatment or vaccine. HIV mutates very often, and it has many subtypes and variants, which makes developing an effective treatment challenging. Therefore, it is important to identify mutations that can lead to drug resistance as well as to identify the subtypes. Studying the evolutionary patterns of HIV is also crucial to understand where this pathogen comes from and what we can expect from it in the future. To identify Drug Resistant Mutations (DRMs), various subtypes, and conduct phylogenetic analysis of sequencing data, various bioinformatic tools and machine learning methods were employed. A pipeline was constructed by combining different bioinformatic software, which was capable of identifying low-frequency DRMs. For identifying different HIV subtypes and studying phylogenetic and evolutionary patterns, both bioinformatic tools and supervised machine learning methods were employed. Each of the two approaches applied succeeded in identifying subtypes and studying phylogenetic relationships, but the feature selection techniques in machine learning used for discovering evolutionary patterns had some limitations. The abundance of sequencing data enables the use of various approaches, such as machine learning, for studying viral genomes. This approach allows for a better understanding of the pathogen and can suggest appropriate solutions for combating it.
  • Post
    Regularized algorithms for applications in microwave thermometry
    (2024) Alsaberi, Rani; Chalmers tekniska högskola / Institutionen för matematiska vetenskaper; Beilina, Larisa; Beilina, Larisa
    This study explores the optimization of the regularization parameter for Tikhonov’s functional, which is derived from the vector wave equation, a partial differential equation (PDE), and transformed into a volume integral equation. Solving this equation using Tikhonov’s functionals provides three different approaches, each with varying numbers of regularization parameters. To identify the optimal regularization parameters, three methods were employed: the L-curve method, the fixed point algorithm, and Morozov’s discrepancy principle. The L-curve method and the fixed point algorithm both indicated that a regularization parameter of 1 is optimal, yielding acceptable results across various reconstruction scenarios. However, Morozov’s discrepancy principle consistently produced superior results, albeit with a significant dependency on the initial reconstruction matrix. Ultimately, while Morozov’s discrepancy principle shows the most promise for optimizing Tikhonov’s functional, it is not yet viable for practical use without further refinement. Moreover, compared to previous studies with a similar setup, this study employs the finite difference method, unlike past works where the finite element method was employed. The code can be found at the link given in [28].
  • Post
    On cosmic censorship for the Einstein-Vlasov system: A numerical study of the spherically symmetric Einstein-Vlasov system for dust-like initial data
    (2024) Martinson, Edvin; Chalmers tekniska högskola / Institutionen för matematiska vetenskaper; Andréasson, Håkan; Andréasson, Håkan
    Already in the first solution of the Einstein field equations of general relativity, singularities appeared. A great deal of effort has since been put into understanding these objects. If there exist singularities from which matter or light could escape, it would lead to a breakdown of the predictability of the Einstein field equations. The concept of cosmic censorship, which asserts that no such singularities exist, is thus fundamental to the theory of general relativity and has been debated in the community ever since it was first conjectured. In this thesis, cosmic censorship is investigated numerically in the spherically symmetric, asymptotically flat Einstein- Vlasov system. The Vlasov matter model is compared to the more primitive dust model, which is known to generate naked singularities that contradict cosmic censorship. The Einstein-Vlasov system is simulated with a Particle In Cell method. The results indicate that initial data that is known to produce naked singularities in dust matter does not give naked singularities in Vlasov matter with large initial velocity dispersion. Moreover, it is shown that the Einstein-Vlasov system can approximate the dust system arbitrarily well up to numerical accuracy. In order to show that cosmic censorship holds even for Vlasov matter that is initially arbitrarily close to dust, better precision in the simulations and a combination of numerical and analytical studies are needed.