| 1. |
- Antonsson, Stefan, 1979-
(författare)
-
Strategies for improving kraftliner pulp properties
- 2008
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- A large part of the world paper manufacturing consists of production of corrugated board components, kraftliner and fluting, that are used in many different types of corrugated boxes. Because these boxes are stored and transported, they are often subjected to changes in relative humidity. These changes together with mechanical loads will increase the deformation of the boxes compared to the case where the same loads are applied in a static environment. This enlarged creep due to the changes in relative humidity is called mechano-sorptive or accelerated creep. Mechano-sorptive creep forces producers to use high safety factors when designing boxes, and therefore, this is one of the key properties of kraftliner boards. Different strategies to decrease mechano-sorptive creep, and to simultaneously gain more knowledge about the causes for this phenomenon in paper, are the aim of this work. Derivatised and underivatised black liquor lignins, a by-product produced in pulp mills in large quantities, have been used together with biomimetic methods, to modify the properties of kraftliner pulp. Furthermore, the properties of kraftliner pulp have been compared to other pulps in order to evaluate the influence of fibre morphological factors, such as fibre width and shape factor, on the mechano-sorptive creep. In addition the influence of the chemical composition of the kraftliner pulp has been evaluated both by means of treating a kraftliner pulp with chlorite and xylanase and by producing pulps with different chemical composition. By using lignin and biomimetic methods, to create radical coupling reactions, it has been shown that it is possible to increase the wet strength of kraftliner pulp sheets. This method of treating the pulp showed, however, no significant effects on the mechano-sorptive creep. The addition of an apolar suberin-like lignin derivative, which has been shown to be possible to produce from natural resources, did show a positive effect on mechano-sorptive creep properties, but at the expense of stiffness properties in constant climate. Different pulps were compared with a kraftliner pulp and it was observed that the ratio between tensile stiffness and hygroexpansion can be used to estimate the mechano-sorptive creep properties. The hardwood kraft pulps investigated had lower hygroexpansion, probably due to more slender and straighter fibres, and higher tensile stiffness, probably due to lower lignin content. As the lignin content was varied by different methods in kraft pulps, it was observed that increased lignin content gives an increased hygroexpansion and decreased tensile stiffness as well as an increased mechano-sorptive creep. There were also indications of increased mechano-sorptive creep due to higher xylan content.
|
|
| 2. |
- Classon, Elisabet, 1965-
(författare)
-
Representing sounds and spellings : Phonological decline and compensatory working memory in acquired hearing impairment
- 2013
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- This thesis examined phonological processing in adults with postlingually acquired moderate-to-severe hearing impairment (HI) and whether explicit working memory processing of phonology and individual working memory capacity (WMC) can compensate for degraded phonological representations in this group (papers I-III). A second aim was to provide reference data for a test of WMC, the reading span test, and to examine the relation between reading span test performance and speech recognition in noise in a larger sample of 50-89 year olds with HI (paper IV). Non-auditory tasks of phonological processing, episodic long-term memory and WMC were used in papers I-III, and both behavioral and electrophysiological measures were collected. Results showed that phonological processing was impaired in the group with HI but that WMC and explicit working memory processing of phonology could be employed to compensate for degraded phonological representations. However, this compensation may come at the cost of interfering with episodic memory encoding. An electrophysiological marker of HI in text-based rhyme judgments was found. Paper IV presented reference data for reading span test performance in two versions of the test in individuals with HI, and results suggesting that WMC may be differentially predictive of speech recognition in noise in different age groups of older adults with HI. The clinical implications of the present results concerns the double disadvantage of individuals with lower WMC and HI. A structured assessment of WMC in rehabilitative settings would help to identify these individuals and tailor treatment to their needs. The reading span test is suggested as a suitable future candidate for clinical WMC assessment.
|
|
| 3. |
- Fay, Dominik
(författare)
-
Towards Scalable Machine Learning with Privacy Protection
- 2023
-
Licentiatavhandling (övrigt vetenskapligt/konstnärligt)abstract
- The increasing size and complexity of datasets have accelerated the development of machine learning models and exposed the need for more scalable solutions. This thesis explores challenges associated with large-scale machine learning under data privacy constraints. With the growth of machine learning models, traditional privacy methods such as data anonymization are becoming insufficient. Thus, we delve into alternative approaches, such as differential privacy.Our research addresses the following core areas in the context of scalable privacy-preserving machine learning: First, we examine the implications of data dimensionality on privacy for the application of medical image analysis. We extend the classification algorithm Private Aggregation of Teacher Ensembles (PATE) to deal with high-dimensional labels, and demonstrate that dimensionality reduction can be used to improve privacy. Second, we consider the impact of hyperparameter selection on privacy. Here, we propose a novel adaptive technique for hyperparameter selection in differentially gradient-based optimization. Third, we investigate sampling-based solutions to scale differentially private machine learning to dataset with a large number of records. We study the privacy-enhancing properties of importance sampling, highlighting that it can outperform uniform sub-sampling not only in terms of sample efficiency but also in terms of privacy.The three techniques developed in this thesis improve the scalability of machine learning while ensuring robust privacy protection, and aim to offer solutions for the effective and safe application of machine learning in large datasets.
|
|
| 4. |
- Sun, Aijun, 1973-
(författare)
-
Radiolabeled acetate PET in oncology imaging : studies on head and neck cancer, prostate cancer and normal distribution
- 2010
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- The use of positron emission tomography (PET) for imaging in oncology has grown rapidly in recent years. 2-[18F]-fluorodeoxyglucose (FDG) is the most common tracer of PET, although drawbacks exist. Radiolabeled 1-[11C]-acetate (C-AC) is a simple probe for evaluation of perfusion, anabolism (lipogenesis) and catabolism (oxidative metabolism) in all living tissues. This study explored the potential of AC PET in head and neck cancer, benign and malignant lymph nodes in prostate cancer and normal distribution. In head and neck cancer, C-AC PET detected more primaries and lymph node metastases than FDG PET. The mean primary tumor volumes delineated by C-AC was 51% larger than that of FDG before radiotherapy (RT). Both FDG and C-AC PET tumor volumes must be carefully validated before used in clinical routine. Baseline tumor clearance rate (kmono) was higher in complete responders (CR) than that in partial responders (PR). kmono tended to correlate inversely with FDG SUV at baseline. Radiosensitive tumors might rely predominantly on oxidative metabolism for their biogenetic needs. kmono increased in PR during RT. The potential reversibility of impaired kmono in radioresistant tumors imply that treatment targeting the intermediary metabolism might improve the outcome. Tumor relative perfusion index (rF) and kmono were coupled in CR throughout the RT, but not in PR. Dynamic C-AC PET provides a new non-invasive method to simultaneously evaluate the tumor oxidative metabolism and perfusion which link the RT response in patients by a single tracer injection. In prostate cancer, elevated C-AC accumulation is common in benign inguinal lymph nodes, probably due to increased lipogenesis rather than lymphatic drainage. CT Hounsfield unit of benign nodes was lower than that of metastases, suggesting that density measurement using CT might improve the specificity of nodal staging of prostate cancer. A novel tracer 2-[18F]-fluoroacetate (F-AC) was synthesized and used for dynamic PET-CT imaging in animals. Compared with C-AC PET-CT, F-AC showed prolonged blood retention, no detectable trapping in myocardium and salivary glands, rapid excretion from liver to bile and urine and de-fluorination resulting in intensive skeletal activity. F-AC does not mimic the normal physiologic path of C-AC and appears to be of little use for assessment of perfusion, intermediary metabolism or lipogenesis.
|
|
| 5. |
- Andersson, Bengt-Åke
(författare)
-
Circulating Biomarkers in Patients with Head and Neck Cancer and the Influence of Cigarette Smoking
- 2019
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Head and neck cancer (HNC) is a collective name for heterogeneous tumors located in the head and neck regions for which smoking, alcohol and human papillomavirus (HPV) are documented risk factors. The survival of HNC patients has only improved marginally during the last decade. The most important prognostic factors are tumor size, local spread and distant metastases, tumor node metastasis (TNM) staging. Prognostic biomarkers are needed as a complement to TNM staging.The aim for this thesis was to investigate rapid and low cost blood based biomarkers which could indicate the risk of HNC, recurrence of the disease or the survival of HNC patients. Furthermore, the aim was to examine how cigarette smoking influences the levels of biomarkers.In paper I, a possible role of plasma cytokines or proteins associated with immune response or inflammation, as biomarkers for the survival of HNC patients was investigated. Higher levels of C-reactive protein (CRP) and tumor necrosis factor alpha (TNF-α) were detected in plasma of the patients compared with the levels in the controls. The elevated levels of these two biomarkers detected in patients were associated with decreased survival.In paper II, the influence of 45 single nucleotide polymorphisms (SNPs) located in 41 genes associated with cell cycle progression, cell death, DNA repair or immune response on cancer risk, tumor recurrence and survival in HNC patients were investigated. SNPs in immune response genes were associated with risk for HNC, an elevated risk for recurrence and a decreased survival in HNC patients.In paper III, the influence of cigarette smoking on levels of inflammatory cells, proteins or cytokines/chemokines, microRNAs (miRNAs) and SNPs was analysed in healthy smokers and non-smokers. Higher levels of total white blood cells (WBCs), neutrophils, monocytes, lymphocytes, neutrophil to lymphocyte ratio (NLR), CRP, monocyte chemoattractant protein- 1 (MCP-1) and interferon gamma (IFN-γ) were detected in smokers compared to non-smokers and indicate an inflammatory response. Also, a lower level of oncomiRNA miR-21was detected in smokers. This alteration, in combination with the elevated levels of IFN-γ in smokers could be a protective response to cigarette smoke. The higher levels of IFN-γ in smokers compared to non-smokers were however only detected in individuals with SNP rs2069705 genotype AG/GG. This indicates a genetic association of the levels of IFN-γ.In paper IV, the separate effects of cigarette smoking and HNC on inflammatory or immune biomarkers and the impact of high risk human papillomavirus, age and gender were investigated. Comparisons of circulating levels of WBCs and its subpopulations, plasma proteins or cytokines/chemokines between smoking and non-smoking patients, smoking and non-smoking controls and between the patient and control groups were analysed. Smoking had highest impact on elevated levels of WBCs, IFN-γ and MCP-1, and HNC had highest impact on elevated levels of neutrophils, monocytes, NLR, CRP, macrophage inflammatory protein 1 beta and TNF-α.In conclusion, host immune response associated parameters could be suitable as biomarkers for the risk of HNC, risk of recurrence or in predicting survival of HNC patients. This thesis show that HNC are associated with systemic inflammatory response and upregulated CRP and TNF-α is related to shorter survival in HNC patients. Additionally, SNPs in immune response genes such as rs1800629 in the TNF-α gene indicates a risk for HNC or an elevated risk for recurrence and a decreased survival in HNC patients. These rapid and low cost blood based biomarkers could be used in combination or as a supplement to established biomarkers in the clinic for a more personalized treatment modality.
|
|
| 6. |
- Ariu, Kaito
(författare)
-
Inference and Online Learning in Structured Stochastic Systems
- 2023
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- This thesis contributes to the field of stochastic online learning problems, with a collection of six papers each addressing unique aspects of online learning and inference problems under specific structures. The first four papers focus on exploration and inference problems, uncovering fundamental information-theoretic limits and efficient algorithms under various structures. The last two papers focus on maximizing rewards by efficiently leveraging these structures.The first paper addresses the complex problem of learning to cluster items based on binary user feedback for multiple questions. It establishes information-theoretical error lower bounds for both uniform and adaptive selection strategies under a fixed budget of rounds or users, and proposes an adaptive algorithm that efficiently allocates the budget.The second paper tackles the challenge of uncovering hidden communities in the Labeled Stochastic Block Model using single-shot observations of labels. It introduces a computationally efficient algorithm, Instance-Adaptive Clustering, which is the first to match instance-specific lower bounds on the expected number of misclassified items.The third paper delves into the best-arm identification or simple regret minimization problem within a Bayesian setting. It takes into consideration a prior distribution for the bandit problem and the expectation of simple regret with respect to that distribution, defining it as Bayesian simple regret.It characterizes the rate of Bayesian simple regret assuming certain continuity conditions on the prior, revealing that the leading term of Bayesian simple regret stems from parameters where the gap between optimal and suboptimal actions is less than . The fourth paper contributes to the fixed budget best-arm identification problem for two-arm bandits with Bernoulli rewards. It demonstrates the optimality of uniform sampling, which evenly samples the arms.It proves that no algorithm can outperform uniform sampling while being at least as good as uniform sampling for some bandit instances.The fifth paper revisits the regret minimization problem in sparse stochastic contextual linear bandits. It introduces a new algorithm, the Thresholded Lasso Bandit, which estimates the linear reward function and its sparse support, and then selects an arm based on these estimations. The algorithm achieves superior regret upper bounds compared to previous algorithms and numerically outperforms them.The sixth and final paper provides a theoretical analysis of recommendation systems in an online setting under unknown user-item preference probabilities and some structures. It derives regret lower bounds based on various structural assumptions and designs optimal algorithms that achieve these bounds. The analysis reveals the relative weights of the different components of regret, providing valuable insights into the efficient algorithms for online recommendation systems.This thesis addresses the technical challenge of structured stochastic online learning problems, providing new insights into the power and limitations of adaptivity in these problems.
|
|
| 7. |
- Arnström, Daniel, 1994-
(författare)
-
On Complexity Certification of Active-Set QP Methods with Applications to Linear MPC
- 2021
-
Licentiatavhandling (övrigt vetenskapligt/konstnärligt)abstract
- In model predictive control (MPC) an optimization problem has to be solved at each time step, which in real-time applications makes it important to solve these efficiently and to have good upper bounds on worst-case solution time. Often for linear MPC problems, the optimization problem in question is a quadratic program (QP) that depends on parameters such as system states and reference signals. A popular class of methods for solving such QPs is active-set methods, where a sequence of linear systems of equations is solved. The primary contribution of this thesis is a method which determines which sequence of subproblems a popular class of such active-set algorithms need to solve, for every possible QP instance that might arise from a given linear MPC problem (i.e, for every possible state and reference signal). By knowing these sequences, worst-case bounds on how many iterations, floating-point operations and, ultimately, the maximum solution time, these active-set algorithms require to compute a solution can be determined, which is of importance when, e.g, linear MPC is used in safety-critical applications. After establishing this complexity certification method, its applicability is extended by showing how it can be used indirectly to certify the complexity of another, efficient, type of active-set QP algorithm which reformulates the QP as a nonnegative least-squares method. Finally, the proposed complexity certification method is extended further to situations when enhancements to the active-set algorithms are used, namely, when they are terminated early (to save computations) and when outer proximal-point iterations are performed (to improve numerical stability).
|
|
| 8. |
- Biel, Martin
(författare)
-
Distributed Stochastic Programming with Applications to Large-Scale Hydropower Operations
- 2021
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Stochastic programming is a subfield of mathematical programming concerned with optimization problems subjected to uncertainty. Many engineering problems with random elements can be accurately modeled as a stochastic program. In particular, decision problems associated with hydropower operations motivate the application of stochastic programming. When complex decision-support problems are considered, the corresponding stochastic programming models often grow too large to store and solve on a single computer. This warrants a need for parallel approaches to enable efficient treatment of large-scale stochastic programs in a distributed environment. In this thesis, we develop mathematical and computational tools to efficiently store and solve distributed stochastic programs. First, we present a software framework for stochastic programming implemented in the Julia programming language. A key feature of the framework is the support for distributing stochastic programs in memory. Moreover, the framework includes a large set of structure-exploiting algorithms for solving stochastic programming problems. These algorithms are based on the classical L-shaped, progressive-hedging, and quasi-gradient algorithms and can be run in parallel on distributed stochastic programs. The distributed performance of our software framework is improved by exploring algorithmic innovations and software patterns. We present the architecture of the framework and highlight key implementation details. Finally, we provide illustrative examples of stochastic programming functionality and benchmarks on large-scale problems. Then, we pursue further algorithmic improvements to the distributed L-shaped algorithm. Specifically, we consider the use of dynamic cut aggregation. We develop theoretical results on convergence and complexity and then showcase performance improvements in numerical experiments. We suggest several aggregation schemes that are based on parameterized selection rules. In brief, cut aggregation can bring major performance improvements to L-shaped algorithms in distributed settings. Next, we consider a fast smoothing scheme for large-scale stochastic programming. We derive a smooth approximation of the subproblems in the quasi-gradient algorithm. This allows us to utilize modern acceleration methods for gradient descent. We derive problem-dependent approximation bounds and convergence properties and note a trade-off between accuracy and speed. We then pose a hybrid procedure that is both fast and accurate and show that it is competitive with the L-shaped method on large-scale benchmarks. Finally, we consider applications to hydropower operations. We consider three case studies in the Swedish river Skellefteälven. The day-ahead planning problem involves specifying optimal order volumes in a deregulated electricity market, without knowledge of the next-day market price, and then optimizing the hydropower production. We provide a detailed introduction to the day-ahead model and explain how it can be implemented in our framework. Using a sample-based algorithm that internally relies on our structure-exploiting solvers, we obtain tight confidence intervals around the optimal solution of the day-ahead problem. We then consider a maintenance scheduling problem as a variation of the day-ahead problem. Last, we consider a capacity expansion problem with a long planning horizon.
|
|
| 9. |
- Blomfeldt, Thomas, 1982-
(författare)
-
Gluten Protein-Based Microcellular Foams and Composites: Development and Functional Properties
- 2010
-
Licentiatavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Most common foams are produced from non-renewable resources (e.g., synthetic plastics),with a number of environmental concerns, hence there is a demand for alternative bio-derivedfoam materials. Wheat gluten protein is widely known to have excellent foaming properties(e.g., in bread making) and is a possible alternative resource for making foam products.Gluten foams were produced using a lyophilization process (freeze-drying) and variousgluten/water-based mixtures were studied. Foams with varying properties were obtained bymixing various amounts of wheat gluten with glycerol (plasticizer) and bacterial cellulosefibers (reinforcement). The gluten foams looked like bread with a beige color and few visuallydetectable surface pores. They were generally characterized as having an open cell structurewith a porosity in the range 75-85% and pore sizes ranging between 20 and 73 μm. Differentmechanical properties were obtained by using varying gluten concentrations and the differentadditives. Plasticizing with glycerol lead to increased flexibility of the foams, with the abilityto recover up to 95% after being compressed by 80%. By reinforcing with bacterial cellulosefibers the material became stiffer, with an increased elastic modulus. Confocal lasermicroscopy revealed that the fibers and gluten interacted. Analyzing the protein structure ofthe foams revealed that the different additives resulted in structures with different proteinpatterns. The samples containing glycerol were more polymerized and less extractable in SDS,whereas the fiber containing samples were only polymerized in small regions and easilyextracted in SDS. Generally the gluten foams had low conductivity values, with some valuesbelow 0.05 W/(m K), which was found to be dependant on density and pore structure. Glutenfoams were also shown to be more difficult to ignite when compared to other conventionalfoams. It was further observed that the foams did not drip making it increasingly difficult forthe fire to spread.
|
|
| 10. |
- Brännmark, Lars-Johan, 1974-
(författare)
-
Robust Sound Field Control for Audio Reproduction : A Polynomial Approach to Discrete-Time Acoustic Modeling and Filter Design
- 2011
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- This thesis is concerned with the design and analysis of robust discrete-time filters for audio equalization and sound field control in real reverberant environments. Inspired by methods in polynomial control theory, a unified framework for acoustic modeling and filter design is developed. The work on modeling is centered around three main themes: First, the acoustic channel between a loudspeaker and a point in space is studied in time, frequency and space, and a polynomial matrix fraction description with diagonal denominator is selected as a physically motivated channel model. As a means for representing channel uncertainties, a probabilistic design model is proposed. Second, the concept of sound field dimensionality, based on the Karhunen-Loève expansion of the sound field, is explored and integrated into the polynomial systems context. Third, a method for spatial interpolation of acoustic transfer functions is proposed and evaluated. Interpolation errors are accounted for by applying the probabilistic uncertainty model to the interpolated data. The work on filter design can be categorized into single- and multichannel methods. The single-channel problem concerns the improvement of the impulse and frequency responses of a single loudspeaker over a region in space, by means of a scalar prefilter. This problem is posed in a SIMO (single-input multiple-output) feedforward control setting, and is solved using polynomial methods. The solution offers several useful insights and results. In particular, new results are derived regarding the adverse pre-ringing problem associated with mixed phase filters. Based on the new results, a refined mixed phase method is proposed that is practically free from pre-ringing artifacts. In the multichannel problem, a desired spatio-temporal sound field is approximated by the joint use of several loudspeakers. This problem is initially formulated and solved by feedforward control over a continuous spatial domain, assuming full knowledge of the spatial field. To obtain a practically feasible design, the control criterion is then spatially discretized, resulting in a standard MIMO(multiple-input multiple-output) linear quadratic feedforward control problem. Since information is generally lost in the discretization process, a robust design based on spatial interpolation and probabilistic error modeling is proposed. The multichannel designs are assessed in an automotive setting, using practical measurements of a nine-channel sound system in a car.
|
|
