Search for dissertations about: "regularization"

Showing result 21 - 25 of 148 swedish dissertations containing the word regularization.

  2. 21. Generalization under Model Mismatch and Distributed Learning

    Author : Martin Hellkvist; Ayca Özcelikkale; Anders Ahlén; Martin Jaggi; Uppsala universitet; []
    Keywords : TEKNIK OCH TEKNOLOGIER; ENGINEERING AND TECHNOLOGY; Machine learning; Signal processing; Generalization error; Training error; Double-descent; Double descent; Distributed learning; Distributed optimization; Learning over networks; Model mismatch; Model misspecification; Fake features; Missing features; linear regression; regularization; Machine learning; Maskininlärning;

    Abstract : Machine learning models are typically configured by minimizing the training error over a given training dataset. On the other hand, the main objective is to obtain models that can generalize, i.e., perform well on data unseen during training. READ MORE

  4. 22. Modelling of Delamination Growth in Composite Structures

    Author : Niklas E. Jansson; Chalmers tekniska högskola; []
    Keywords : TEKNIK OCH TEKNOLOGIER; ENGINEERING AND TECHNOLOGY; large deformation; interface; composite materials; 3D-shell; cohesive zone; regularization; shell element; delamination;

    Abstract : Delaminations constitute an important damage and failure mode in polymeric composite laminates. In order to explicitly incorporate and model the delamination progression within a finite element analysis, the meso-modelling concept is applied in the present work. The laminate is thereby represented by separate layers connected with interfaces. READ MORE

  5. 23. Improvement and Assessment of Two-Dimensional Resistivity Models Derived from Radiomagnetotelluric and Direct-Current Resistivity Data

    Author : Thomas Kalscheuer; Laust B. Pedersen; Christopher Juhlin; Roland Roberts; Heinrich Brasse; Uppsala universitet; []
    Keywords : NATURVETENSKAP; NATURAL SCIENCES; Electromagnetic geophysics; radiomagnetotellurics; direct-current resistivity; displacement currents; dielectric effect; non-linearity; inversion; regularization; two-dimensional model; model variance; model resolution; data resolution; Geophysics; Geofysik;

    Abstract : Two-dimensional (2-D) models of electrical resistivity are improved by jointly inverting radiomagnetotelluric (RMT) and direct-current resistivity (DCR) data or by allowing for displacement currents in the inversion of RMT data collected on highly resistive bedrock. Uniqueness and stability of the 2-D models are assessed with a model variance and resolution analysis that allows for the non-linearity of the inverse problem. READ MORE

  6. 24. Land Mine Detection using Dual-Band Electro-Optical Sensing

    Author : Magnus Lundberg; Chalmers tekniska högskola; []
    Keywords : TEKNIK OCH TEKNOLOGIER; ENGINEERING AND TECHNOLOGY; TEKNIK OCH TEKNOLOGIER; ENGINEERING AND TECHNOLOGY; land mines; regularization; nuisance parameters; detection; infrared imaging; occlusion modeling; multi-spectral imaging; prior knowledge; Signalbehandling;

    Abstract : This thesis addresses the detection of buried and partially occluded land mines using electro-optical sensing. According to the Red Cross approximately 10000 people are killed and another 30000 are injured each year in mine related accidents. READ MORE

  7. 25. Sobolev-Type Spaces : Properties of Newtonian Functions Based on Quasi-Banach Function Lattices in Metric Spaces

    Author : Lukáš Malý; Anders Björn; Jana Björn; Tomas Sjödin; Andrea Cianchi; Linköpings universitet; []
    Keywords : NATURVETENSKAP; NATURAL SCIENCES; Newtonian space; Sobolev-type space; metric measure space; upper gradient; Sobolev capacity; Banach function lattice; quasi-normed space; rearrangement-invariant space; maximal operator; Lipschitz function; regularization; weak boundedness; density of Lipschitz functions; quasi-continuity; continuity; doubling measure; Poincaré inequality;

    Abstract : This thesis consists of four papers and focuses on function spaces related to first-order analysis in abstract metric measure spaces. The classical (i.e., Sobolev) theory in Euclidean spaces makes use of summability of distributional gradients, whose definition depends on the linear structure of Rn. READ MORE