Change search
Refine search result
45678 301 - 350 of 392
CiteExportLink to result list
Permanent link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Rows per page
  • 5
  • 10
  • 20
  • 50
  • 100
  • 250
Sort
  • Standard (Relevance)
  • Author A-Ö
  • Author Ö-A
  • Title A-Ö
  • Title Ö-A
  • Publication type A-Ö
  • Publication type Ö-A
  • Issued (Oldest first)
  • Issued (Newest first)
  • Created (Oldest first)
  • Created (Newest first)
  • Last updated (Oldest first)
  • Last updated (Newest first)
  • Disputation date (earliest first)
  • Disputation date (latest first)
  • Standard (Relevance)
  • Author A-Ö
  • Author Ö-A
  • Title A-Ö
  • Title Ö-A
  • Publication type A-Ö
  • Publication type Ö-A
  • Issued (Oldest first)
  • Issued (Newest first)
  • Created (Oldest first)
  • Created (Newest first)
  • Last updated (Oldest first)
  • Last updated (Newest first)
  • Disputation date (earliest first)
  • Disputation date (latest first)
Select
The maximal number of hits you can export is 250. When you want to export more records please use the Create feeds function.
  • 301.
    Pavlenko, Tatjana
    et al.
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Rios, Felix Leopoldo
    Graphical posterior predictive classifier:  Bayesian model averaging with particle GibbsManuscript (preprint) (Other academic)
    Abstract [en]

    In this study, we present a multi-class graphical Bayesian predictive classifier that incorporates the uncertainty in the model selection into the standard Bayesian formalism. For each class, the dependence structure underlying the observed features is represented by a set of decomposable Gaussian graphical models. Emphasis is then placed on the Bayesian model averaging which takes full account of the class-specific model uncertainty by averaging over the posterior graph model probabilities. An explicit evaluation of the model probabilities is well known to be infeasible. To address this issue, we consider the particle Gibbs strategy of Olsson et al. (2016) for posterior sampling from decomposable graphical models which utilizes the Christmas tree algorithm of Olsson et al. (2017) as proposal kernel. We also derive a strong hyper Markov law which we call the hyper normal Wishart law that allow to perform the resultant Bayesian calculations locally. The proposed predictive graphical classifier reveals superior performance compared to the ordinary Bayesian predictive rule that does not account for the model uncertainty, as well as to a number of out-of-the-box classifiers.

  • 302.
    Perninge, Magnus
    et al.
    Department of Automatic Control, Lund University.
    Söder, Lennart
    KTH, School of Electrical Engineering (EES), Electric Power Systems.
    Irreversible Investments with Delayed Reaction: An Application to Generation Re-Dispatch in Power System Operation2014In: Mathematical Methods of Operations Research, ISSN 1432-2994, E-ISSN 1432-5217, Vol. 79, no 2, p. 195-224Article in journal (Refereed)
    Abstract [en]

    In this article we consider how the operator of an electric power system should activate bids on the regulating power market in order to minimize the expected operation cost. Important characteristics of the problem are reaction times of actors on the regulating market and ramp-rates for production changes in power plants. Neglecting these will in general lead to major underestimation of the operation cost. Including reaction times and ramp-rates leads to an impulse control problem with delayed reaction. Two numerical schemes to solve this problem are proposed. The first scheme is based on the least-squares Monte Carlo method developed by Longstaff and Schwartz (Rev Financ Stud 14:113-148, 2001). The second scheme which turns out to be more efficient when solving problems with delays, is based on the regression Monte Carlo method developed by Tsitsiklis and van Roy (IEEE Trans Autom Control 44(10):1840-1851, 1999) and (IEEE Trans Neural Netw 12(4):694-703, 2001). The main contribution of the article is the idea of using stochastic control to find an optimal strategy for power system operation and the numerical solution schemes proposed to solve impulse control problems with delayed reaction.

  • 303.
    Philip, Johan
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.).
    The Area of a Random Convex Polygon2004Report (Other academic)
    Abstract [en]

    We consider the area of the convex hull of n random points in a square. We give the distribution function of thearea for three and four points. We also present some results on the number of vertices of the convex hull. Results from Monte Carlo tests with large n are presented and compared with asymptotic estimates.

  • 304.
    Philip, Johan
    KTH, School of Engineering Sciences (SCI).
    The area of a random triangle in a regular hexagon2010Report (Other academic)
    Abstract [en]

    We determine the distribution function for the area of a random triangle in a regular hexagon.

  • 305.
    Philip, Johan
    KTH, School of Engineering Sciences (SCI).
    The area of a random triangle in a regular pentagon and the golden ratio2012Report (Other academic)
    Abstract [en]

    We determine the distribution function for the area of a random triangle in a regular pentagon. It turns out that the golden ratio is intimately related to the pentagon calculations.

  • 306.
    Philip, Johan
    KTH, School of Engineering Sciences (SCI).
    The area of a random triangle in a square2010Report (Other academic)
    Abstract [en]

    We determine the distribution function for the area of a random triangle in a unit square. The reault is not new. The method presented here is worked out to shed more light on the problem.

  • 307.
    Pokorny, Florian T.
    et al.
    KTH, School of Computer Science and Communication (CSC), Computer Vision and Active Perception, CVAP.
    Ek, Carl Henrik
    KTH, School of Computer Science and Communication (CSC), Computer Vision and Active Perception, CVAP.
    Kjellström, Hedvig
    KTH, School of Computer Science and Communication (CSC), Computer Vision and Active Perception, CVAP.
    Kragic, Danica
    KTH, School of Computer Science and Communication (CSC), Computer Vision and Active Perception, CVAP.
    Persistent Homology for Learning Densities with Bounded Support2012In: Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012 / [ed] P. Bartlett, F.C.N. Pereira, C.J.C. Burges, L. Bottou and K.Q. Weinberger, Curran Associates, Inc., 2012, p. 1817-1825Conference paper (Refereed)
    Abstract [en]

    We present a novel method for learning densities with bounded support which enables us to incorporate 'hard' topological constraints. In particular, we show how emerging techniques from computational algebraic topology and the notion of persistent homology can be combined with kernel-based methods from machine learning for the purpose of density estimation. The proposed formalism facilitates learning of models with bounded support in a principled way, and - by incorporating persistent homology techniques in our approach - we are able to encode algebraic-topological constraints which are not addressed in current state of the art probabilistic models. We study the behaviour of our method on two synthetic examples for various sample sizes and exemplify the benefits of the proposed approach on a real-world dataset by learning a motion model for a race car. We show how to learn a model which respects the underlying topological structure of the racetrack, constraining the trajectories of the car.

  • 308.
    Pokorny, Florian T.
    et al.
    KTH, School of Computer Science and Communication (CSC), Computer Vision and Active Perception, CVAP.
    Ek, Carl Henrik
    KTH, School of Computer Science and Communication (CSC), Computer Vision and Active Perception, CVAP.
    Kjellström, Hedvig
    KTH, School of Computer Science and Communication (CSC), Computer Vision and Active Perception, CVAP.
    Kragic, Danica
    KTH, School of Computer Science and Communication (CSC), Computer Vision and Active Perception, CVAP.
    Topological Constraints and Kernel-Based Density Estimation2012Conference paper (Refereed)
    Abstract [en]

    This extended abstract1 explores the question of how to estimate a probability distribution from a finite number of samples when information about the topology of the support region of an underlying density is known. This workshop contribution is a continuation of our recent work [1] combining persistent homology and kernel-based density estimation for the first time and in which we explored an approach capable of incorporating topological constraints in bandwidth selection. We report on some recent experiments with high-dimensional motion capture data which show that our method is applicable even in high dimensions and develop our ideas for potential future applications of this framework.

  • 309.
    Prevost, Quentin
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Measurement and valuation of country risk: how to get a right value?2016Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The purpose of this master thesis is to focus on country risk and its quantification as a premium. Country risk is an important parameter for investors willing to invest abroad and especially in emerging countries. Indeed, there is additional risk to invest in such countries for numerous reasons. It is thus imperative to be able to quantify it. The actual state of the art about this topic is still at its beginning.

    In this master thesis, I have developed two axis of reflection to get a country risk premium. The first one derives from the Capital Asset Pricing Model and related corporate finance theory. The second axis is based on a more mathematical approach.

    In the end, I have managed to have a quantified results with those two approaches. They are converging for both methods.

    I have applied my results with case studies on two countries: Sweden and Mexico

  • 310.
    Pärlstrand, Erik
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Comparing fast- and slow-acting features for short-term price predictions2017Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    This thesis compares two groups of features for short-term price predictions of futures contracts; fast- and slow-acting features. The fast-acting group are based on limit order book derived features and technical indicators that reacts to changes in price quickly. The slow-acting features constitute of technical indicators that reacts to changes in price slowly.

    The comparison is done through two methods, group importance and a mean cost calculation. This is evaluated for different forecast horizons and contracts. Furthermore, two years of data was provided to do the analysis. Moreover, the comparison is modelled with an ensemble method called random forest. The response is constructed using rolling quantiles and a volume weighted price. 

    The finding implies that fast-acting features are superior at predicting price changes on smaller time scales, while long-acting features are better at predicting prices changes on larger time scales. Furthermore, the multivariate model results were similar to the univariate ones. However, the results are not clear-cut and more investigation ought to be done in order to confirm these results.

  • 311. Radhakrishnan, A.
    et al.
    Solus, Liam
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.).
    Uhler, C.
    Counting Markov equivalence classes by number of immoralities2017In: Uncertainty in Artificial Intelligence - Proceedings of the 33rd Conference, UAI 2017, AUAI Press Corvallis , 2017Conference paper (Refereed)
    Abstract [en]

    Two directed acyclic graphs (DAGs) are called Markov equivalent if and only if they have the same underlying undirected graph (i.e. skeleton) and the same set of immoralities. When using observational data alone and typical identifiability assumptions, such as faithfulness, a DAG model can only be determined up to Markov equivalence. Therefore, it is desirable to understand the size and number of Markov equivalence classes (MECs) combinatorially. In this paper, we address this enumerative question using a pair of generating functions that encode the number and size of MECs on a skeleton G, and in doing so we connect this problem to classical problems in combinatorial optimization. The first generating function is a graph polynomial that counts the number of MECs on G by their number of immoralities. Using connections to the independent set problem, we show that computing a DAG on G with the maximum possible number of immoralities is NP-hard. The second generating function counts the MECs on G according to their size. Via computer enumeration, we show that this generating function is distinct for every connected graph on p nodes for all p < 10.

  • 312. Redjil, Amel
    et al.
    Choutri, Salah eddine
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    On relaxed stochastic optimal control for stochastic differential equations driven by G-Brownian motion2018In: Latin American Journal of Probability and Mathematical Statistics, ISSN 1980-0436, E-ISSN 1980-0436, Vol. 15, no 1, p. 201-212Article in journal (Refereed)
    Abstract [en]

    In the G-framework, we establish existence of an optimal stochastic relaxed control for stochastic differential equations driven by a G-Brownian motion.

  • 313.
    Rehn, Rasmus
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Stochastic modeling of yield curve shifts usingfunctional data analysis2014Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    This thesis approaches the problem of modeling the multivariate distribution of interest rates by implementing a novel tool of statistics known as functional data analysis (FDA). This is done by viewing yield curve shifts as distinct but continuous stochastic objects defined over a continuum of maturities. Based on these techniques, we provide two stochastic models with different assumptions regarding the temporal dependence of yield curve shifts and compare their performance with empirical data. The study finds that both models replicate the distributions of yield changes with medium- and long-term maturities, whereas none of the models perform satisfactory at the short segment of the yield curve. Both models, however, appear to accurately capture the cross-sectional dependence.

  • 314.
    Ringh, Emil
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Optimization and Systems Theory.
    Low complexity algorithms for faster-than-Nyquistsign: Using coding to avoid an NP-hard problem2013Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    This thesis is an investigation of what happens when communication links are pushed towards their limits and the data-bearing-pulses are packed tighter in time than previously done. This is called faster-than-Nyquist (FTN) signaling and it will violate the Nyquist inter-symbol interference criterion, implying that the data-pulsesare no longer orthogonal and thus that the samples at the receiver will be dependent on more than one of the transmitted symbols. Inter-symbol interference (ISI) has occurred and the consequences of it are studied for the AWGN-channel model. Here it is shown that in order to do maximum likelihood estimation on these samples the receiver will face an NP-hard problem. The standard algorithm to make good estimations in the ISI case is the Viterbi algorithm, but applied on a block with N bits and interference among K bits thecomplexity is O(N *2K), hence limiting the practical applicability. Here, a precoding scheme is proposed together with a decoding that reduce the estimation complexity. By applying the proposed precoding/decoding to a data block of length N the estimation can be done in O(N2) operations preceded by a single off-line O(N3) calculation. The precoding itself is also done in O(N2)operations, with a single o ff-line operation of O(N3) complexity.

    The strength of the precoding is shown in simulations. In the first it was tested together with turbo codes of code rate 2/3 and block lengthof 6000 bits. When sending 25% more data (FTN) the non-precoded case needed about 2.5 dB higher signal-to-noise ratio (SNR) to have the same error rate as the precoded case. When the precoded case performed without any block errors, the non-precoded case still had a block error rate almost equal to 1.

    We also studied the scenario of transmission with low latency and high reliability. Here, 600 bits were transmitted with a code rate of 2/3, and hence the target was to communicate 400 bits of data. Applying FTN with doublepacking, that is transmitting 1200 bits during the same amount of time, it was possible to lower the code rate to 1/3 since only 400 bits of data was to be communicated. This technique greatly improves the robustness. When the FTN case performed error free, the classical Nyquist case still had a block error rate of 0.19. To reach error free performance the Nyquist case needed 1.25 dB higher SNR compared to the precoded FTN case with lower code rate.

  • 315.
    Rios, Felix Leopold
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.).
    Personalized health care: Switching to a subpopulation in Phase III2012Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Abstract

    Since different patients may have different causes of getting a disease, treating every patient having a certain disease in the same manner is not always be the best way to go. A treatment having effect in one type of patients may not have the same effect in a different type of patients. This makes it possible to partition a patient population into subpopulations in which a drug has distinct expected response. In this thesis the patient population is partitioned into two subpopulations where we have prior knowledge that one of them has a higher expected response to a drug than the other. Based on responses to a drug in Phase II, it has been analyzed in which of the populations Phase III should continue. The results show that the decision is highly dependent on the utility function on which the analysis is based. One interesting case is when the vast majority of the patient population belongs to the subpopulation with the higher expected response and a utility function that takes into account the prevalence of the populations. In that case the simulations show that when the difference in expected response between the subpopulations is large, it is a safer choice in continuing in Phase III in the subpopulation having the higher expected response than in the full population even though the expected utility will be less. This is an expected result which indicates that the approach used to model the situation studied in this report is reasonable

  • 316.
    Rios, Felix Leopoldo
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Bayesian inference in probabilistic graphical models2017Doctoral thesis, comprehensive summary (Other academic)
    Abstract [en]

    This thesis consists of four papers studying structure learning and Bayesian inference in probabilistic graphical models for both undirected and directed acyclic graphs (DAGs).

    Paper A presents a novel algorithm, called the Christmas tree algorithm (CTA), that incrementally construct junction trees for decomposable graphs by adding one node at a time to the underlying graph. We prove that CTA with positive probability is able to generate all junction trees of any given number of underlying nodes. Importantly for practical applications, we show that the transition probability of the CTA kernel has a computationally tractable expression. Applications of the CTA transition kernel are demonstrated in a sequential Monte Carlo (SMC) setting for counting the number of decomposable graphs.

    Paper B presents the SMC scheme in a more general setting specifically designed for approximating distributions over decomposable graphs. The transition kernel from CTA from Paper A is incorporated as proposal kernel. To improve the traditional SMC algorithm, a particle Gibbs sampler with a systematic refreshment step is further proposed. A simulation study is performed for approximate graph posterior inference within both log-linear and decomposable Gaussian graphical models showing efficiency of the suggested methodology in both cases.

    Paper C explores the particle Gibbs sampling scheme of Paper B for approximate posterior computations in the Bayesian predictive classification framework. Specifically, Bayesian model averaging (BMA) based on the posterior exploration of the class-specific model is incorporated into the predictive classifier to take full account of the model uncertainty. For each class, the dependence structure underlying the observed features is represented by a distribution over the space of decomposable graphs. Due to the intractability of explicit expression, averaging over the approximated graph posterior is performed. The proposed BMA classifier reveals superior performance compared to the ordinary Bayesian predictive classifier that does not account for the model uncertainty, as well as to a number of out-of-the-box classifiers.

    Paper D develops a novel prior distribution over DAGs with the ability to express prior knowledge in terms of graph layerings. In conjunction with the prior, a stochastic optimization algorithm based on the layering property of DAGs is developed for performing structure learning in Bayesian networks. A simulation study shows that the algorithm along with the prior has superior performance compared with existing priors when used for learning graph with a clearly layered structure.

  • 317.
    Rios, Felix Leopoldo
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Bayesian structure learning in graphical models2016Licentiate thesis, comprehensive summary (Other academic)
    Abstract [en]

    This thesis consists of two papers studying structure learning in probabilistic graphical models for both undirected graphs anddirected acyclic graphs (DAGs).

    Paper A, presents a novel family of graph theoretical algorithms, called the junction tree expanders, that incrementally construct junction trees for decomposable graphs. Due to its Markovian property, the junction tree expanders are shown to be suitable for proposal kernels in a sequential Monte Carlo (SMC) sampling scheme for approximating a graph posterior distribution. A simulation study is performed for the case of Gaussian decomposable graphical models showing efficiency of the suggested unified approach for both structural and parametric Bayesian inference.

    Paper B, develops a novel prior distribution over DAGs with the ability to express prior knowledge in terms of graph layerings. In conjunction with the prior, a search and score algorithm based on the layering property of DAGs, is developed for performing structure learning in Bayesian networks. A simulation study shows that the search and score algorithm along with the prior has superior performance for learning graph with a clearly layered structure compared with other priors.

  • 318. Roueff, Francois
    et al.
    Rydén, Tobias
    Lund University.
    Non-parametric estimation of mixing densities for discrete distributions2005In: Annals of Statistics, ISSN 0090-5364, E-ISSN 2168-8966, Vol. 33, no 5, p. 2066-2108Article in journal (Refereed)
    Abstract [en]

    By a mixture density is meant a density of the form πμ(⋅)=∫πθ(⋅)×μ(dθ), where (πθ)θ∈Θ is a family of probability densities and μ is a probability measure on Θ. We consider the problem of identifying the unknown part of this model, the mixing distribution μ, from a finite sample of independent observations from πμ. Assuming that the mixing distribution has a density function, we wish to estimate this density within appropriate function classes. A general approach is proposed and its scope of application is investigated in the case of discrete distributions. Mixtures of power series distributions are more specifically studied. Standard methods for density estimation, such as kernel estimators, are available in this context, and it has been shown that these methods are rate optimal or almost rate optimal in balls of various smoothness spaces. For instance, these results apply to mixtures of the Poisson distribution parameterized by its mean. Estimators based on orthogonal polynomial sequences have also been proposed and shown to achieve similar rates. The general approach of this paper extends and simplifies such results. For instance, it allows us to prove asymptotic minimax efficiency over certain smoothness classes of the above-mentioned polynomial estimator in the Poisson case. We also study discrete location mixtures, or discrete deconvolution, and mixtures of discrete uniform distributions.

  • 319. Rubenthaler, Sylvain
    et al.
    Rydén, Tobias
    Lund University.
    Wiktorsson, Magnus
    Fast simulated annealing in Rd with an application to maximum likelihood estimation in state-space models2009In: Stochastic Processes and their Applications, ISSN 0304-4149, E-ISSN 1879-209X, Vol. 119, no 6, p. 1912-1931Article in journal (Refereed)
    Abstract [en]

    We study simulated annealing algorithms to maximise a function psi on a subset of R(d). In classical simulated annealing, given a current state theta(n) in stage n of the algorithm, the probability to accept a proposed state z at which psi is smaller, is exp(-beta(n+1)(psi(z) - psi (theta(n))) where (beta(n)) is the inverse temperature. With the standard logarithmic increase of (beta(n)) the probability P(psi(theta(n)) <= psi(max) - epsilon), with psi(max) the maximal value of psi, then tends to zero at a logarithmic rate as n increases. We examine variations of this scheme in which (beta(n)) is allowed to grow faster, but also consider other functions than the exponential for determining acceptance probabilities. The main result shows that faster rates of convergence can be obtained, both with the exponential and other acceptance functions. We also show how the algorithm may be applied to functions that cannot be computed exactly but only approximated, and give an example of maximising the log-likelihood function for a state-space model.

  • 320.
    Rydén, Otto
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Statistical learning procedures for analysis of residential property price indexes2017Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Residential Price Property Indexes (RPPIs) are used to study the price development of residential property over time. Modeling and analysing an RPPI is not straightforward due to residential property being a heterogeneous good. This thesis focuses on analysing the properties of the two most conventional hedonic index modeling approaches, the hedonic time dummy method and the hedonic imputation method. These two methods are analysed with statistical learning procedures from a regression perspective, specifically, ordinary least squares regression, and a number of more advanced regression approaches, Huber regression, lasso regression, ridge regression and principal component regression. The analysis is based on the data from 56 000 apartment transactions in Stockholm during the period 2013-2016 and results in several models of a RPPI. These suggested models are then validated using both qualitative and quantitative methods, specifically a bootstrap re-sampling to perform analyses of an empirical confidence interval for the index values and a mean squared errors analysis of the different index periods. Main results of this thesis show that the hedonic time dummy index methodology produces indexes with smaller variances and more robust indexes for smaller datasets. It is further shown that modeling of RPPIs with robust regression generally results in a more stable index that is less affected by outliers in the underlying transaction data. This type of robust regression strategy is therefore recommended for a commercial implementation of an RPPI.

  • 321.
    Rydén, Tobias
    Lund University.
    EM versus Markov chain Monte Carlo for estimation of hidden Markov models: a computational perspective2008In: Bayesian Analysis, ISSN 1931-6690, Vol. 3, no 4, p. 659-688Article in journal (Refereed)
    Abstract [en]

    Hidden Markov models (HMMs) and related models have become standard in statistics during the last 15-20 years, with applications in diverse areas like speech and other statistical signal processing, hydrology, financial statistics and econometrics, bioinformatics etc. Inference in HMMs is traditionally often carried out using the EM algorithm, but examples of Bayesian estimation, in general implemented through Markov chain Monte Carlo (MCMC) sampling are also frequent in the HMM literature. The purpose of this paper is to compare the EM and MCMC approaches in three cases of different complexity; the examples include model order selection, continuous-time HMMs and variants of HMMs in which the observed data depends on many hidden variables in an overlapping fashion. All these examples in some way or another originate from real-data applications. Neither EM nor MCMC analysis of HMMs is a black-box methodology without need for user-interaction, and we will illustrate some of the problems, like poor mixing and long computation times, one may expect to encounter.

  • 322.
    Rydén, Tobias
    Lund University.
    Hidden Markov Models2004In: Encyclopedia of Actuarial Science: vol 2 / [ed] Teugels, J., and Sundt, B., Wiley-Blackwell, 2004, p. 821-827Chapter in book (Refereed)
  • 323.
    Röhss, Josefine
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    A Statistical Framework for Classification of Tumor Type from microRNA Data2016Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Hepatocellular carcinoma (HCC) is a type of liver cancer with low survival rate, not least due to the difficulty of diagnosing it in an early stage. The objective of this thesis is to build a random forest classification method based on microRNA (and messenger RNA) expression profiles from patients with HCC. The main purpose is to be able to distinguish between tumor samples and normal samples by measuring the miRNA expression. If successful, this method can be used to detect HCC at an earlier stage and to design new therapeutics. The microRNAs and messenger RNAs which have a significant difference in expression between tumor samples and normal samples are selected for building random forest classification models. These models are then tested on paired samples of tumor and surrounding normal tissue from patients with HCC. The results show that the classification models built for classifying tumor and normal samples have high prediction accuracy and hence show high potential for using microRNA and messenger RNA expression levels for diagnosis of HCC.

  • 324.
    Samani, Forough Shahab
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS), Network and Systems engineering.
    Stadler, Rolf
    KTH, School of Electrical Engineering and Computer Science (EECS), Network and Systems engineering. KTH Royal Inst Technol, Dept Network & Syst Engn, Stockholm, Sweden..
    Predicting Distributions of Service Metrics using Neural Networks2018In: 2018 14TH INTERNATIONAL CONFERENCE ON NETWORK AND SERVICE MANAGEMENT (CNSM) / [ed] Salsano, S Riggio, R Ahmed, T Samak, T DosSantos, CRP, IEEE , 2018, p. 45-53Conference paper (Refereed)
    Abstract [en]

    We predict the conditional distributions of service metrics, such as response time or frame rate, from infrastructure measurements in a cloud environment. From such distributions, key statistics of the service metrics, including mean, variance, or percentiles can be computed, which are essential for predicting SLA conformance or enabling service assurance. We model the distributions as Gaussian mixtures, whose parameters we predict using mixture density networks, a class of neural networks. We apply the method to a Voll service and a KY store running on our lab testbed. The results validate the effectiveness of the method when applied to operational data. In the case of predicting the mean of the frame rate or response time, the accuracy matches that of random forest, a baseline model.

  • 325. Seita, D.
    et al.
    Pokorny, Florian T.
    KTH, School of Computer Science and Communication (CSC), Computer Vision and Active Perception, CVAP. KTH, School of Computer Science and Communication (CSC), Centres, Centre for Autonomous Systems, CAS.
    Mahler, J.
    Kragic, Danica
    KTH, School of Computer Science and Communication (CSC), Centres, Centre for Autonomous Systems, CAS. KTH, School of Computer Science and Communication (CSC), Robotics, perception and learning, RPL.
    Franklin, M.
    Canny, J.
    Goldberg, K.
    Large-scale supervised learning of the grasp robustness of surface patch pairs2017In: 2016 IEEE International Conference on Simulation, Modeling, and Programming for Autonomous Robots, SIMPAR 2016, Institute of Electrical and Electronics Engineers Inc. , 2017, p. 216-223Conference paper (Refereed)
    Abstract [en]

    The robustness of a parallel-jaw grasp can be estimated by Monte Carlo sampling of perturbations in pose and friction but this is not computationally efficient. As an alternative, we consider fast methods using large-scale supervised learning, where the input is a description of a local surface patch at each of two contact points. We train and test with disjoint subsets of a corpus of 1.66 million grasps where robustness is estimated by Monte Carlo sampling using Dex-Net 1.0. We use the BIDMach machine learning toolkit to compare the performance of two supervised learning methods: Random Forests and Deep Learning. We find that both of these methods learn to estimate grasp robustness fairly reliably in terms of Mean Absolute Error (MAE) and ROC Area Under Curve (AUC) on a held-out test set. Speedups over Monte Carlo sampling are approximately 7500x for Random Forests and 1500x for Deep Learning.

  • 326.
    Serpeka, Rokas
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.).
    Analyzing and modelling exchange rate data using VAR framework2012Independent thesis Advanced level (degree of Master (One Year)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Abstract

     

    In this report analysis of foreign exchange rates time series are performed. First, triangular arbitrage is detected and eliminated from data series using linear algebra tools. Then Vector Autoregressive processes are calibrated and used to replicate dynamics of exchange rates as well as to forecast time series. Finally, optimal portfolio of currencies with minimal Expected Shortfall is formed using one time period ahead forecasts

  • 327.
    Shahrabi Farahani, Hossein
    KTH, School of Computer Science and Communication (CSC), Computational Biology, CB.
    Computational Modeling of Cancer Progression2013Doctoral thesis, comprehensive summary (Other academic)
    Abstract [en]

    Cancer is a multi-stage process resulting from accumulation of genetic mutations. Data obtained from assaying a tumor only contains the set of mutations in the tumor and lacks information about their temporal order. Learning the chronological order of the genetic mutations is an important step towards understanding the disease. The probability of introduction of a mutation to a tumor increases if certain mutations that promote it, already happened. Such dependencies induce what we call the monotonicity property in cancer progression. A realistic model of cancer progression should take this property into account.

    In this thesis, we present two models for cancer progression and algorithms for learning them. In the first model, we propose Progression Networks (PNs), which are a special class of Bayesian networks. In learning PNs the issue of monotonicity is taken into consideration. The problem of learning PNs is reduced to Mixed Integer Linear Programming (MILP), which is a NP-hard problem for which very good heuristics exist. We also developed a program, DiProg, for learning PNs.

    In the second model, the problem of noise in the biological experiments is addressed by introducing hidden variable. We call this model Hidden variable Oncogenetic Network (HON). In a HON, there are two variables assigned to each node, a hidden variable that represents the progression of cancer to the node and an observable random variable that represents the observation of the mutation corresponding to the node. We devised a structural Expectation Maximization (EM) algorithm for learning HONs. In the M-step of the structural EM algorithm, we need to perform a considerable number of inference tasks. Because exact inference is tractable only on Bayesian networks with bounded treewidth, we also developed an algorithm for learning bounded treewidth Bayesian networks by reducing the problem to a MILP.

    Our algorithms performed well on synthetic data. We also tested them on cytogenetic data from renal cell carcinoma. The learned progression networks from both algorithms are in agreement with the previously published results.

    MicroRNAs are short non-coding RNAs that are involved in post transcriptional regulation. A-to-I editing of microRNAs converts adenosine to inosine in the double stranded RNA. We developed a method for determining editing levels in mature microRNAs from the high-throughput RNA sequencing data from the mouse brain. Here, for the first time, we showed that the level of editing increases with development. 

  • 328.
    Shahrabi Farahani, Hossein
    et al.
    KTH, School of Computer Science and Communication (CSC), Computational Biology, CB.
    Lagergren, Jens
    KTH, School of Computer Science and Communication (CSC), Computational Biology, CB.
    A structural EM algorithm for learning hidden variable oncogenetic networksManuscript (preprint) (Other academic)
  • 329.
    Shahrabi Farahani, Hossein
    et al.
    KTH, School of Computer Science and Communication (CSC), Computational Biology, CB.
    Parviainen, Pekka
    KTH, School of Computer Science and Communication (CSC), Computational Biology, CB.
    Lagergren, Jens
    KTH, School of Computer Science and Communication (CSC), Computational Biology, CB.
    A Linear Programming Approach for Learning Bounded Treewidth Bayesian Networks2013Manuscript (preprint) (Other academic)
    Abstract [en]

    In many applications, one wants to compute conditional probabilities from a Bayesian network. This inference problem is NP-hard in general but becomes tractable when the network has bounded treewidth. Motivated by the needs of applications, we study learning bounded treewidth Bayesian networks. We formulate this problem as a mixed integer linear program (MILP) which can be solved by an anytime algorithm. 

  • 330. Shi, Guodong
    et al.
    Proutiere, Alexandre
    KTH, School of Electrical Engineering (EES), Automatic Control. KTH, School of Electrical Engineering (EES), Centres, ACCESS Linnaeus Centre.
    Johansson, Mikael
    KTH, School of Electrical Engineering (EES), Automatic Control. KTH, School of Electrical Engineering (EES), Centres, ACCESS Linnaeus Centre.
    Baras, John S.
    Johansson, Karl H.
    KTH, School of Electrical Engineering (EES), Automatic Control. KTH, School of Electrical Engineering (EES), Centres, ACCESS Linnaeus Centre.
    The Evolution of Beliefs over Signed Social Networks2016In: Operations Research, ISSN 0030-364X, E-ISSN 1526-5463, Vol. 64, no 3, p. 585-604Article in journal (Refereed)
    Abstract [en]

    We study the evolution of opinions (or beliefs) over a social network modeled as a signed graph. The sign attached to an edge in this graph characterizes whether the corresponding individuals or end nodes are friends (positive links) or enemies (negative links). Pairs of nodes are randomly selected to interact over time, and when two nodes interact, each of them updates its opinion based on the opinion of the other node and the sign of the corresponding link. This model generalizes the DeGroot model to account for negative links: when two adversaries interact, their opinions go in opposite directions. We provide conditions for convergence and divergence in expectation, in mean-square, and in almost sure sense and exhibit phase transition phenomena for these notions of convergence depending on the parameters of the opinion update model and on the structure of the underlying graph. We establish a no-survivor theorem, stating that the difference in opinions of any two nodes diverges whenever opinions in the network diverge as a whole. We also prove a live-or-die lemma, indicating that almost surely, the opinions either converge to an agreement or diverge. Finally, we extend our analysis to cases where opinions have hard lower and upper limits. In these cases, we study when and how opinions may become asymptotically clustered to the belief boundaries and highlight the crucial influence of (strong or weak) structural balance of the underlying network on this clustering phenomenon.

  • 331.
    Singh, Alex
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    A risk-transaction cost trade-off model for index tracking2014Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    This master thesis considers and evaluates a few different risk models for stock portfolios, including an ordinary sample covariance matrix, factor models and an approach inspired from random matrix theory. The risk models are evaluated by simulating minimum variance portfolios and employing a cross-validation. The Bloomberg+ transaction cost model is investigated and used to optimize portfolios of stocks, with respect to a trade off between the active risk of the portfolio and transaction costs. Further a few different simulations are performed while using the optimizer to rebalance long-only portfolios. The optimization problem is solved using an active-set algorithm. A couple of approaches are shown that may be used to visually try to decide a value for the risk aversion parameter λ in the objective function of the optimization problem.

    The thesis concludes that there is a practical difference between the different risk models that are evaluated. The ordinary sample covariance matrix is shown to not perform as well as the other models. It also shows that more frequent rebalancing is preferable to less frequent. Further the thesis goes on to show a peculiar behavior of the optimization problem, which is that the optimizer does not rebalance all the way to 0 in simulations, even if enough time is provided, unless it is explicitly required by the constraints.

  • 332.
    Singh, Ravi Shankar
    et al.
    KTH, School of Electrical Engineering and Computer Science (EECS), Electric Power and Energy Systems.
    Hooshyar, Hossein
    KTH, School of Electrical Engineering and Computer Science (EECS), Electric Power and Energy Systems.
    Vanfretti, Luigi
    KTH.
    Experimental Real-Time Testing of a Decentralized PMU Data-Based Power Systems Mode Estimator2017In: 2017 IEEE POWER & ENERGY SOCIETY GENERAL MEETING, IEEE , 2017Conference paper (Refereed)
    Abstract [en]

    This paper presents the results and testing of a Phasor Measurement Unit (PMU) data-based mode estimation application deployed within a decentralized architecture using a real-time test platform. This work is a continuation of that in [1], which described a decentralized mode estimation architecture that enables the application to better detect local modes whose observability is affected by other more observable modes. The tests in this paper were carried out using an active distribution network (ADN) comprised of a high voltage network connected to a distribution grid including renewable energy resources (RES). The developed application was run in a decentralized architecture where each PMU was associated with its own processing unit which was running the application to estimate modes from the time-series data. The results of the decentralized mode estimation architecture are analyzed and compared with its centralized counterpart.

  • 333.
    Singull, Martin
    et al.
    Linköpings universitet .
    Koski, Timo
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    On the Distribution of Matrix Quadratic Forms2012In: Communications in Statistics - Theory and Methods, ISSN 0361-0926, E-ISSN 1532-415X, Vol. 41, no 18, p. 3403-3415Article in journal (Refereed)
    Abstract [en]

     A characterization of the distribution of the multivariate quadratic form given by XAX', where X is a p x n normally distributed matrix and A is an n x n symmetric real matrix, is presented. We show that the distribution of the quadratic form is the same as the distribution of a weighted sum of non central Wishart distributed matrices. This is applied to derive the distribution of the sample covariance between the rows of X when the expectation is the same for every column and is estimated with the regular mean.

  • 334.
    Sjöberg, Lars Erik
    KTH, School of Architecture and the Built Environment (ABE), Urban Planning and Environment, Geoinformatik och Geodesi.
    On the Best Quadratic Minimum Bias Non-Negative Estimator of a Two-Variance Component Model2011In: Journal of Geodetic Science, ISSN 2081-9943, Vol. 1, no 3, p. 280-285Article in journal (Refereed)
    Abstract [en]

    Variance components (VCs) in linear adjustment models are usually successfully computed by unbiased estimators. However, for many unbiased VC techniques estimated variance components might be negative, a result that cannot be tolerated by the user. This is, for example, the case with the simple additive VC model aσ2/1 + bσ2/2 with known coefficients a and b, where either of the unbiasedly estimated variance components σ2/1 + σ2/2 may frequently come out negative. This fact calls for so-called non-negative VC estimators. Here the Best Quadratic Minimum Bias Non-negative Estimator (BQMBNE) of a two-variance component model is derived. A special case with independent observations is explicitly presented.

  • 335.
    Skanke, Björn
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Analysis of Pension Strategies2014Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    In a time where people tend to retire earlier and live longer in combination with an augmented personal responsibility of allocating or at least choosing adequately composed pension funds, the importance of a deeper understanding of long term investment strategies is inevitably accentuated. On the background of discrepancies in suggested pension fund strategies by influential fund providers, professional advisers and previous literature, this thesis aims at addressing foremost one particular research question: How should an investor optimally allocate between risky and risk-less assets in a pension fund depending on age? In order to answer the question the sum of Human wealth, defined as the present value of all expected future incomes, and ordinary Financial wealth is maximized by applying a mean-variance and a expected utility approach. The latter, and mathematically more sound method yields a strategy suggesting 100% of available capital to be invested in risky assets until the age of 47 whereafter the portion should be gradually reduced and reach the level of 32% at the last period before retirement. The strategy is clearly favorable to solely holding a risk-free asset and it just outperforms the commonly applied "100 minus age"-strategy.

  • 336. Sköld, Martin
    et al.
    Rydén, Tobias
    Lund University.
    Samuelsson, Viktoria
    Bratt, Charlotte
    Ekblad, Lars
    Olsson, Håkan
    Baldetorp, Bo
    Regression analysis and modelling of data acquisition for SELDI-TOF mass spectrometry2007In: Bioinformatics, ISSN 1367-4803, E-ISSN 1367-4811, Vol. 23, no 11, p. 1401-1409Article in journal (Refereed)
    Abstract [en]

    Motivation: Pre-processing of SELDI-TOF mass spectrometry data is currently performed on a largel y ad hoc basis. This makes comparison of results from independent analyses troublesome and does not provide a framework for distinguishing different sources of variation in data. Results: In this article, we consider the task of pooling a large number of single-shot spectra, a task commonly performed automatically by the instrument software. By viewing the underlying statistical problem as one of heteroscedastic linear regression, we provide a framework for introducing robust methods and for dealing with missing data resulting from a limited span of recordable intensity values provided by the instrument. Our framework provides an interpretation of currently used methods as a maximum-likelihood estimator and allows theoretical derivation of its variance. We observe that this variance depends crucially on the total number of ionic species, which can vary considerably between different pooled spectra. This variation in variance can potentially invalidate the results from naive methods of discrimination/classification and we outline appropriate data transformations. Introducing methods from robust statistics did not improve the standard errors of the pooled samples. Imputing missing values however-using the EM algorithm-had a notable effect on the result; for our data, the pooled height of peaks which were frequently truncated increased by up to 30%.

  • 337.
    Stattin, Oskar
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Large scale inference under sparse and weak alternatives: non-asymptotic phase diagram for CsCsHM statistics2017Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    High-throughput measurement technology allows to generate and store huge amounts of features, of which very few can be useful for any one single problem at hand. Examples include genomics, proteomics and astronomy, where massive multiple testing often needs to be per- formed, expecting a few significant effects and essentially a null back- ground. A number of new test procedures have been developed for detecting these, so-called sparse and weak effects, in large scale statistical inference. The most widely used is Higher Criticism, HC (see e.g. Donoho and Jin (2004)). A new class of goodness-of-fit test statistics, called CsCsHM, has recently been derived (see Stepanova and Pavlenko (2017)) for the same type of multiple testing, it is shown to achieve better asymptotic properties than the traditional HC approach.This report empirically investigates the behavior of both test procedures in the neighborhood of the detection boundary, i.e. the threshold for the detectability of sparse and weak effects. This theoretical boundary sharply separates the phase space, spanned by the sparsity and weakness parameters, into two subregions the region of detectability and the region of undetectability. The statistics are also applied and compared for both methodologies for features selection in high dimensional binary classification problems. Besides the study of the methods and simulations, applications of both methods on realistic data are carried out. It is found that the statistics are comparable in performance accuracy. 

  • 338.
    Steffen, Richard
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Risk premia implied by derivative prices2015Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The thesis investigates the potential to recover the real world probabilities of an underlying asset from derivative prices by using the recovery approach developed in (Carr & Yu, 2012) and (Ross, 2011). For this purpose the VIX Index and US Treasury bills are used to recover the VIX dynamics and the short rate dynamics under the real world probability measure. The approach implies that VIX and its derivatives has a risk premium equal to zero contradicting empirical evidence of a substantial negative risk premium. In fact, we show that for any asset unrelated to the short rate its risk premium is zero. In the case of recovering the short rate, the CIR model is calibrated to the US zero coupon Treasury yield curve. The predictions of the recovered CIR process is benchmarked against the risk neutral CIR process and a naive predictor. The recovered process is found to outperform the risk neutral process suggesting that the recovery step was successful. However, it underperforms the naive process in its predictions.

  • 339.
    Stenberg, Kristoffer
    et al.
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Wikerman, Henrik
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Evaluating Regime Switching in Dynamic Conditional Correlation2013Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    This paper provides a comparative study of the Dynamic Conditional Correlation model introduced by Engle (2002) and the Independent Switching Dynamic Conditional Correlation model introduced by Lee (2010) by evaluating the models for a set of known correlation processes. The evaluation is also extended to cover empirical data to assess the practical performance of the models. The data include the price of gold and oil, the yield on benchmark 10 year U.S. Treasury notes and the Euro-U.S. dollar exchange rate from January 2007 to December 2009. In addition, a general description of the difficulties of estimating correlations is presented to give the reader a better understanding of the limitations of the models. From the results, it is concluded that there is no general superiority of neither the IS-DCC model nor the DCC model, except for very short-lived correlation shifts. For short-lived shifts, the IS-DCC model outperforms in both detecting and measuring correlations. However, this paper recommends that these models are used in combination with a qualitative study in empirical situations to better understand the underlying correlation dynamics.

  • 340. Stepanova, N.
    et al.
    Pavlenko, Tatjana
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Goodness-of-fit tests based on sup-functionals of weighted empirical processes2018In: Theory of Probability and its Applications, ISSN 0040-585X, E-ISSN 1095-7219, Vol. 63, no 2, p. 292-317Article in journal (Refereed)
    Abstract [en]

    A large class of goodness-of-fit test statistics based on sup-functionals of weighted empirical processes is proposed and studied. The weight functions employed are the Erdős–Feller– Kolmogorov–Petrovski upper-class functions of a Brownian bridge. Based on the result of M. Csörgő, S. Csörgő, L. Horváth, and D. Mason on this type of test statistics, we provide the asymptotic null distribution theory for the class of tests and present an algorithm for tabulating the limit distribution functions under the null hypothesis. A new family of nonparametric confidence bands is constructed for the true distribution function and is found to perform very well. The results obtained, involving a new result on the convergence in distribution of the higher criticism statistic, as introduced by D. Donoho and J. Jin, demonstrate the advantage of our approach over a common approach that utilizes a family of regularly varying weight functions. Furthermore, we show that, in various subtle problems of detecting sparse heterogeneous mixtures, the proposed test statistics achieve the detection boundary found by Yu. I. Ingster and, when distinguishing between the null and alternative hypotheses, perform optimally adaptively to unknown sparsity and size of the non-null effects. 

  • 341. Stjernqvist, Susann
    et al.
    Rydén, Tobias
    Lund University.
    A continuous-index hidden Markov jump process for modeling DNA copy number data2009In: Biostatistics, ISSN 1465-4644, E-ISSN 1468-4357, Vol. 10, no 4, p. 773-778Article in journal (Refereed)
    Abstract [en]

    The number of copies of DNA in human cells can be measured using array comparative genomic hybridization (aCGH), which provides intensity ratios of sample to reference DNA at genomic locations corresponding to probes on a microarray. In the present paper, we devise a statistical model, based on a latent continuous-index Markov jump process, that is aimed to capture certain features of aCGH data, including probes that are unevenly long, unevenly spaced, and overlapping. The model has a continuous state space, with 1 state representing a normal copy number of 2, and the rest of the states being either amplifications or deletions. We adopt a Bayesian approach and apply Markov chain Monte Carlo (MCMC) methods for estimating the parameters and the Markov process. The model can be applied to data from both tiling bacterial artificial chromosome arrays and oligonucleotide arrays. We also compare a model with normal distributed noise to a model with t-distributed noise, showing that the latter is more robust to outliers.

  • 342. Stjernqvist, Susann
    et al.
    Rydén, Tobias
    Lund University.
    Sköld, Martin
    Staaf, Johan
    Continuous-index hidden Markov modelling of array CGH copy number data2007In: Bioinformatics, ISSN 1367-4803, E-ISSN 1367-4811, Vol. 23, no 8, p. 1006-1014Article in journal (Refereed)
  • 343.
    Styrud, Lovisa
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Risk Premium Prediction of Car Damage Insurance using Artificial Neural Networks and Generalized Linear Models2017Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Over the last few years the interest in statistical learning methods, in particular artificial neural networks, has reawakened due to increasing computing capacity, available data and a strive towards automatization of different tasks. Artificial neural networks have numerous applications, why they appear in various contexts. Using artificial neural networks in insurance rate making is an area in which a few pioneering studies have been conducted, with promising results. This thesis suggests using a multilayer perceptron neural network for pricing car damage insurance. The MLP is compared with two traditionally used methods within the framework of generalized linear models. The MLP was selected by cross-validation of a set of candidate models. For the comparison models, a log-link GLM with Tweedie's compound Poisson distribution modeling the risk premium as dependent variable was set up, as well as a two-parted GLM with a log-link Poisson GLM for claim frequency and a log-link Gamma GLM for claim severity. Predictions on an independent test set showed that the Tweedie GLM had the lowest prediction error, followed by the MLP model and last the Poisson-Gamma GLM. Analysis of risk ratios for the different explanatory variables showed that the Tweedie GLM was also the least discriminatory model, followed by the Poisson-Gamma GLM and the MLP. The MLP had the highest bootstrap estimate of variance in prediction error on the test set. Overall however, the MLP model performed roughly in line with the GLM models and given the basic model configurations cross-validated and the restricted computing power, the MLP results should be seen as successful for the use of artificial neural networks in car damage insurance rate making. Nevertheless, practical aspects argue in favor of using GLM.

    This thesis is written at If P&C Insurance, a property and casualty insurance company active in Scandinavia, Finland and the Baltic countries. The headquarters are situated in Bergshamra, Stockholm.

  • 344.
    Su, Xun
    et al.
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Cheung, Mei Ting
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Day-of-the-week eects in stock market data2012Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    The purpose of this thesis is to investigate day-of-the-week effects for stock index returns. The investigations include analysis of means and variances as well as return-distribution properties such as skewness and tail behavior. Moreover, the existences of conditional day-of-the-week effects, depending on the outcome of returns from the previous week, are analyzed. Particular emphasis is put on determining useful testing procedures for differences in variance in return data from different weekdays. Two time series models, AR and GARCH(1,1), are used to find out if any weekday's mean return is different from other days. The investigations are repeated for two-day re- turns and for returns of diversified portfolios made up of several stock index returns.

  • 345.
    Sundberg, Victor
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Application and Bootstrapping of the Munich Chain Ladder Method2016Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    Point estimates of the Standard Chain Ladder method (CLM) and of the more complex Munich Chain Ladder method (MCL) are compared to real data on 38 different datasets in order to evaluate if MCL produces better predictions on average with a dataset from an arbitrary insurance portfolio. MCL is also examined to determine if the future paid and incurred claims converge as time progresses. A bootstrap model based on MCL (BMCL) is examined in order to evaluate its possibility to estimate the probability density function (PDF) of future claims and observable claim development results (OCDR). The results show that the paid and incurred predictions by MCL converge. The results also show that when considering all datasets MCL produce on average better estimations than CLM with paid data but no improvement can be seen with incurred data. Further the results show that by considering a subset of datasets which fulfil certain criteria, or by only considering accident years after 1999 the percentage of datasets in which MCL produce superior estimations increases. When examining BMCL one finds that it can produce estimated PDFs of ultimate reserves and OCDRs, however the mean of estimate of ultimate reserves does not converge to the MCL estimates nor do the mean of the OCDRs converge to zero. In order to get the right convergence the estimated OCDR PDFs are centered and the mean of the BMCL estimated ultimate reserve is set to the MCL estimate by multiplication.

  • 346.
    Sundgren, David
    et al.
    Univ Gavle, Gavle, Sweden..
    Ekenberg, Love
    KTH.
    Danielson, Mats
    Univ Stockholm, S-10691 Stockholm, Sweden..
    Shifted Dirichlet Distributions as Second-Order Probability Distributions that Factors into Marginals2009In: ISIPTA '09: PROCEEDINGS OF THE SIXTH INTERNATIONAL SYMPOSIUM ON IMPRECISE PROBABILITY: THEORIES AND APPLICATIONS / [ed] Augustin, T Coolen, FPA Moral, S Troffaes, MCM, SOC IMPRECISE PROBABILITY THEORY & APPLICATIONS-SIPTA , 2009, p. 405-+Conference paper (Refereed)
    Abstract [en]

    In classic decision theory it is assumed that a decision-maker can assign precise numerical values corresponding to the true value of each consequence, as well as precise numerical probabilities for their occurrences. In attempting to address real-life problems, where uncertainty in the input data prevails, some kind of representation of imprecise information is important. Second-order distributions, probability distributions over probabilities, is one way to achieve such a representation. However, it is hard to intuitively understand statements in a multi-dimensional space and user statements must be provided more locally. But the information-theoretic interplay between joint and marginal distributions may give rise to unwanted effects on the global level. We consider this problem in a setting of second-order probability distributions and find a family of distributions that normalised over the probability simplex equals its own product of marginals. For such distributions, there is no flow of information between the joint distributions and the marginal distributions other than the trivial fact that the variables belong to the probability simplex.

  • 347.
    Sundin, Jesper
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Risk contribution and its application in asset and risk management for life insurance2016Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

    In risk management one important aspect is the allocation of total portfolio risk into its components. This can be done by measuring each components' risk contribution relative to the total risk, taking into account the covariance between components. The measurement procedure is straightforward under assumptions of elliptical distributions but not under the commonly used multivariate log-normal distributions. Two portfolio strategies are considered, the "buy and hold" and the "constant mix" strategy. The profits and losses of the components of a generic portfolio strategy are defined in order to enable a proper definition of risk contribution for the constant mix strategy. Then kernel estimation of risk contribution is performed for both portfolio strategies using Monte Carlo simulation. Further, applications for asset and risk management with risk contributions are discussed in the context of life insurance.

  • 348.
    Sundqvist, Greger
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.).
    Model risk in a hedging perspective2011Independent thesis Advanced level (degree of Master (One Year)), 20 credits / 30 HE creditsStudent thesis
  • 349.
    Svensson Depraetere, Xavier
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
    Application of new particle-based solutions to the Simultaneous Localization and Mapping (SLAM) problem2017Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
    Abstract [en]

        In this thesis, we explore novel solutions to the Simultaneous Localization and Mapping (SLAM) problem based on particle filtering and smoothing methods. In essence, the SLAM problem constitutes of two interdependent tasks: map building and tracking. Three solution methods utilizing different smoothing techniques are explored. The smoothing methods used are fixed lag smoothing (FLS), forward-only forward-filtering backward-smoothing (forward-only FFBSm) and the particle-based, rapid incremental smoother (PaRIS). In conjunction with these smoothing techniques the well-established Expectation-Maximization (EM) algorithm is used to produce maximum-likelihood estimates of the map. The three solution method are then evaluated and compared in a simulated setting.

  • 350.
    Svensson, Jens
    KTH, School of Engineering Sciences (SCI), Mathematics (Dept.).
    On Importance Sampling and Dependence Modeling2009Doctoral thesis, comprehensive summary (Other academic)
    Abstract [en]

    This thesis consists of four papers.

    In the first paper, Monte Carlo simulation for tail probabilities of heavy-tailed random walks is considered. Importance sampling algorithms are constructed by using mixtures of the original distribution with some other state-dependent distributions. Sufficient conditions under which the relative error of such algorithms is bounded are found, and the bound is calculated. A new mixture algorithm based on scaling of the original distribution is presented and compared to existing algorithms.

    In the second paper, Monte Carlo simulation of quantiles is treated. It is shown that by using importance sampling algorithms developed for tail probability estimation, efficient quantile estimators can be obtained. A functional limit of the quantile process under the importance sampling measure is found, and the variance of the limit process is calculated for regularly varying distributions. The procedure is also applied to the calculation of expected shortfall. The algorithms are illustrated numerically for a heavy-tailed random walk.

    In the third paper, large deviation probabilities for a sum of dependent random variables are derived. The dependence stems from a few underlying random variables, so-called factors. Each summand is composed of two parts: an idiosyncratic part and a part given by the factors. Conditions under which both factors and idiosyncratic components contribute to the large deviation behavior are found, and the resulting approximation is evaluated in a simple example.

    In the fourth paper, the asymptotic eigenvalue distribution of the exponentially weighted moving average covariance estimator is studied. Equations for the asymptotic spectral density and the boundaries of its support are found using the Marchenko-Pastur theorem.

45678 301 - 350 of 392
CiteExportLink to result list
Permanent link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf