Data fusion and matching by maximizing statistical dependencies

The core aim of machine learning is to make a computer program learn from the experience. Learning from data is usually defined as a task of learning regularities or patterns in data in order to extract useful information, or to learn the underlying concept. An important sub-field of machine learning is called multi-view learning where the task is to learn from multiple data sets or views describing the same underlying concept. A typical example of such scenario would be to study a biological concept using several biological measurements like gene expression, protein expression and metabolic profiles, or to classify web pages based on their content and the contents of their hyperlinks. In this thesis, novel problem formulations and methods for multi-view learning are presented. The contributions include a linear data fusion approach during exploratory data analysis, a new measure to evaluate different kinds of representations for textual data, and an extension of multi-view learning for novel scenarios where the correspondence of samples in the different views or data sets is not known in advance. In order to infer the one-to-one correspondence of samples between two views, a novel concept of multi-view matching is proposed. The matching algorithm is completely data-driven and is demonstrated in several applications such as matching of metabolites between humans and mice, and matching of sentences between documents in two languages.

A Probabilistic Approach to the Primary Visual Cortex

What can the statistical structure of natural images teach us about the human brain? Even though the visual cortex is one of the most studied parts of the brain, surprisingly little is known about how exactly images are processed to leave us with a coherent percept of the world around us, so we can recognize a friend or drive on a crowded street without any effort. By constructing probabilistic models of natural images, the goal of this thesis is to understand the structure of the stimulus that is the raison d etre for the visual system. Following the hypothesis that the optimal processing has to be matched to the structure of that stimulus, we attempt to derive computational principles, features that the visual system should compute, and properties that cells in the visual system should have. Starting from machine learning techniques such as principal component analysis and independent component analysis we construct a variety of sta- tistical models to discover structure in natural images that can be linked to receptive field properties of neurons in primary visual cortex such as simple and complex cells. We show that by representing images with phase invariant, complex cell-like units, a better statistical description of the vi- sual environment is obtained than with linear simple cell units, and that complex cell pooling can be learned by estimating both layers of a two-layer model of natural images. We investigate how a simplified model of the processing in the retina, where adaptation and contrast normalization take place, is connected to the nat- ural stimulus statistics. Analyzing the effect that retinal gain control has on later cortical processing, we propose a novel method to perform gain control in a data-driven way. Finally we show how models like those pre- sented here can be extended to capture whole visual scenes rather than just small image patches. By using a Markov random field approach we can model images of arbitrary size, while still being able to estimate the model parameters from the data.

Kantoaalto koulutuksesta työyhteisön viestintäosaamiseen : lyhytkestoisen viestintäkoulutuksen vaikuttavuuden avaintekijät

The purpose of this research is to examine whether short-term communication training can have an impact on the improvement of communication capacity of working communities, and what are prerequisites for the creation of such capacity. Subjects of this research were short-term communication trainings aimed at the managerial and expert levels of enterprises and communities. The research endeavors to find out how communication trainings with an impact should be devised and implemented, and what this requires from the client and provider of the training service. The research data is mostly comprised of quantitative feed-back collected at the end of a training day, as well as delayed interviews. The evaluations have been based on a stakeholder approach, and those concerned were participants to the trainings, clients having commissioned the trainings and communication trainers. The principal method of the qualitative analysis is that of a data-driven content analysis. Two research instruments have been constructed for the analysis and for the presentation of the results: an evaluation circle for the purposes of a holistic evaluation and a development matrix for the structuring of an effective training. The core concept of the matrix is a carrier wave effect, which is needed to carry the abstractions from the training into concrete functions in the everyday life. The relevance of the results has been tested in a pilot organization. The immediate assessment and delayed evaluations gave a very differing picture of the trainings. The immediate feedback was of nearly commendable level, but the effects carried forward into the everyday situations of the working community were small and that the learning rarely was applied into practice. A training session that receives good feedback does not automatically result in the development of individual competence, let alone that of the community. The results show that even short-term communication training can promote communication competence that eventually changes the working culture on an organizational level, provided that the training is designed into a process and that the connections into the participants’ work are ensured. It is essential that all eight elements of the carrier wave effect are taken into account. The entire purchaser-provider -process must function while not omitting the contribution of the participants themselves. The research illustrates the so called bow tie -model of an effective communication training based on the carrier wave effect. Testing the results in pilot trainings showed that a rather small change in the training approach may have a signi¬ficant effect on the outcome of the training as well as those effects that are carried on into the working community. The evaluation circle proved to be a useful tool, which can be used while planning, executing and evaluating training in practice. The development matrix works as a tool for those producing the training service, those using the service as well as those deciding on the purchase of the service in planning and evaluating training that sustainably improves communication capacity. Thus the evaluation circle also works to support and ensure the long-term effects of short-term trainings. In addition to communication trainings, the tools developed for this research are useable for many such needs, where an organization is looking to improve its operations and profitability through training.

Modeling Asymmetries in Financial Data with Multiplicative Error Models

This thesis addresses modeling of financial time series, especially stock market returns and daily price ranges. Modeling data of this kind can be approached with so-called multiplicative error models (MEM). These models nest several well known time series models such as GARCH, ACD and CARR models. They are able to capture many well established features of financial time series including volatility clustering and leptokurtosis. In contrast to these phenomena, different kinds of asymmetries have received relatively little attention in the existing literature. In this thesis asymmetries arise from various sources. They are observed in both conditional and unconditional distributions, for variables with non-negative values and for variables that have values on the real line. In the multivariate context asymmetries can be observed in the marginal distributions as well as in the relationships of the variables modeled. New methods for all these cases are proposed. Chapter 2 considers GARCH models and modeling of returns of two stock market indices. The chapter introduces the so-called generalized hyperbolic (GH) GARCH model to account for asymmetries in both conditional and unconditional distribution. In particular, two special cases of the GARCH-GH model which describe the data most accurately are proposed. They are found to improve the fit of the model when compared to symmetric GARCH models. The advantages of accounting for asymmetries are also observed through Value-at-Risk applications. Both theoretical and empirical contributions are provided in Chapter 3 of the thesis. In this chapter the so-called mixture conditional autoregressive range (MCARR) model is introduced, examined and applied to daily price ranges of the Hang Seng Index. The conditions for the strict and weak stationarity of the model as well as an expression for the autocorrelation function are obtained by writing the MCARR model as a first order autoregressive process with random coefficients. The chapter also introduces inverse gamma (IG) distribution to CARR models. The advantages of CARR-IG and MCARR-IG specifications over conventional CARR models are found in the empirical application both in- and out-of-sample. Chapter 4 discusses the simultaneous modeling of absolute returns and daily price ranges. In this part of the thesis a vector multiplicative error model (VMEM) with asymmetric Gumbel copula is found to provide substantial benefits over the existing VMEM models based on elliptical copulas. The proposed specification is able to capture the highly asymmetric dependence of the modeled variables thereby improving the performance of the model considerably. The economic significance of the results obtained is established when the information content of the volatility forecasts derived is examined.

Modeling of Surface UV Radiation Using Satellite Data

Solar UV radiation is harmful for life on planet Earth, but fortunately the atmospheric oxygen and ozone absorb almost entirely the most energetic UVC radiation photons. However, part of the UVB radiation and much of the UVA radiation reaches the surface of the Earth, and affect human health, environment, materials and drive atmospheric and aquatic photochemical processes. In order to quantify these effects and processes there is a need for ground-based UV measurements and radiative transfer modeling to estimate the amounts of UV radiation reaching the biosphere. Satellite measurements with their near-global spatial coverage and long-term data conti-nuity offer an attractive option for estimation of the surface UV radiation. This work focuses on radiative transfer theory based methods used for estimation of the UV radiation reaching the surface of the Earth. The objectives of the thesis were to implement the surface UV algorithm originally developed at NASA Goddard Space Flight Center for estimation of the surface UV irradiance from the meas-urements of the Dutch-Finnish built Ozone Monitoring Instrument (OMI), to improve the original surface UV algorithm especially in relation with snow cover, to validate the OMI-derived daily surface UV doses against ground-based measurements, and to demonstrate how the satellite-derived surface UV data can be used to study the effects of the UV radiation. The thesis consists of seven original papers and a summary. The summary includes an introduction of the OMI instrument, a review of the methods used for modeling of the surface UV using satellite data as well as the con-clusions of the main results of the original papers. The first two papers describe the algorithm used for estimation of the surface UV amounts from the OMI measurements as well as the unique Very Fast Delivery processing system developed for processing of the OMI data received at the Sodankylä satellite data centre. The third and the fourth papers present algorithm improvements related to the surface UV albedo of the snow-covered land. Fifth paper presents the results of the comparison of the OMI-derived daily erythemal doses with those calculated from the ground-based measurement data. It gives an estimate of the expected accuracy of the OMI-derived sur-face UV doses for various atmospheric and other conditions, and discusses the causes of the differences between the satellite-derived and ground-based data. The last two papers demonstrate the use of the satellite-derived sur-face UV data. Sixth paper presents an assessment of the photochemical decomposition rates in aquatic environment. Seventh paper presents use of satellite-derived daily surface UV doses for planning of the outdoor material weathering tests.

Owner-driven decision support in holding-specific forest planning

Socio-economic and demographic changes among family forest owners and demands for versatile forestry decision aid motivated this study, which sought grounds for owner-driven forest planning. Finnish family forest owners’ forest-related decision making was analyzed in two interview-based qualitative studies, the main findings of which were surveyed quantitatively. Thereafter, a scheme for adaptively mixing methods in individually tailored decision support processes was constructed. The first study assessed owners’ decision-making strategies by examining varying levels of the sharing of decision-making power and the desire to learn. Five decision-making modes – trusting, learning, managing, pondering, and decisive – were discerned and discussed against conformable decision-aid approaches. The second study conceptualized smooth communication and assessed emotional, practical, and institutional boosters of and barriers to such smoothness in communicative decision support. The results emphasize the roles of trust, comprehension, and contextual services in owners’ communicative decision making. In the third study, a questionnaire tool to measure owners’ attitudes towards communicative planning was constructed by using trusting, learning, and decisive dimensions. Through a multivariate analysis of survey data, three owner groups were identified as fusions of the original decision-making modes: trusting learners (53%), decisive learners (27%), and decisive managers (20%). Differently weighted communicative services are recommended for these compound wishes. The findings of the studies above were synthesized in a form of adaptive decision analysis (ADA), which allows and encourages the decision-maker (owner) to make deliberate choices concerning the phases of a decision aid (planning) process. The ADA model relies on adaptability and feedback management, which foster smooth communication with the owner and (inter-)organizational learning of the planning institution(s). The summarized results indicate that recognizing the communication-related amenity values of family forest owners may be crucial in developing planning and extension services. It is therefore recommended that owners, root-level planners, consultation professionals, and pragmatic researchers collaboratively continue to seek stable change.

On probability-based inference under data missing by design

Whether a statistician wants to complement a probability model for observed data with a prior distribution and carry out fully probabilistic inference, or base the inference only on the likelihood function, may be a fundamental question in theory, but in practice it may well be of less importance if the likelihood contains much more information than the prior. Maximum likelihood inference can be justified as a Gaussian approximation at the posterior mode, using flat priors. However, in situations where parametric assumptions in standard statistical models would be too rigid, more flexible model formulation, combined with fully probabilistic inference, can be achieved using hierarchical Bayesian parametrization. This work includes five articles, all of which apply probability modeling under various problems involving incomplete observation. Three of the papers apply maximum likelihood estimation and two of them hierarchical Bayesian modeling. Because maximum likelihood may be presented as a special case of Bayesian inference, but not the other way round, in the introductory part of this work we present a framework for probability-based inference using only Bayesian concepts. We also re-derive some results presented in the original articles using the toolbox equipped herein, to show that they are also justifiable under this more general framework. Here the assumption of exchangeability and de Finetti's representation theorem are applied repeatedly for justifying the use of standard parametric probability models with conditionally independent likelihood contributions. It is argued that this same reasoning can be applied also under sampling from a finite population. The main emphasis here is in probability-based inference under incomplete observation due to study design. This is illustrated using a generic two-phase cohort sampling design as an example. The alternative approaches presented for analysis of such a design are full likelihood, which utilizes all observed information, and conditional likelihood, which is restricted to a completely observed set, conditioning on the rule that generated that set. Conditional likelihood inference is also applied for a joint analysis of prevalence and incidence data, a situation subject to both left censoring and left truncation. Other topics covered are model uncertainty and causal inference using posterior predictive distributions. We formulate a non-parametric monotonic regression model for one or more covariates and a Bayesian estimation procedure, and apply the model in the context of optimal sequential treatment regimes, demonstrating that inference based on posterior predictive distributions is feasible also in this case.

Discovering hidden structures in molecular data using a Bayesian partition model approach

Advancements in the analysis techniques have led to a rapid accumulation of biological data in databases. Such data often are in the form of sequences of observations, examples including DNA sequences and amino acid sequences of proteins. The scale and quality of the data give promises of answering various biologically relevant questions in more detail than what has been possible before. For example, one may wish to identify areas in an amino acid sequence, which are important for the function of the corresponding protein, or investigate how characteristics on the level of DNA sequence affect the adaptation of a bacterial species to its environment. Many of the interesting questions are intimately associated with the understanding of the evolutionary relationships among the items under consideration. The aim of this work is to develop novel statistical models and computational techniques to meet with the challenge of deriving meaning from the increasing amounts of data. Our main concern is on modeling the evolutionary relationships based on the observed molecular data. We operate within a Bayesian statistical framework, which allows a probabilistic quantification of the uncertainties related to a particular solution. As the basis of our modeling approach we utilize a partition model, which is used to describe the structure of data by appropriately dividing the data items into clusters of related items. Generalizations and modifications of the partition model are developed and applied to various problems. Large-scale data sets provide also a computational challenge. The models used to describe the data must be realistic enough to capture the essential features of the current modeling task but, at the same time, simple enough to make it possible to carry out the inference in practice. The partition model fulfills these two requirements. The problem-specific features can be taken into account by modifying the prior probability distributions of the model parameters. The computational efficiency stems from the ability to integrate out the parameters of the partition model analytically, which enables the use of efficient stochastic search algorithms.

Computationally Efficient Methods for MDL-Optimal Density Estimation and Data Clustering

The Minimum Description Length (MDL) principle is a general, well-founded theoretical formalization of statistical modeling. The most important notion of MDL is the stochastic complexity, which can be interpreted as the shortest description length of a given sample of data relative to a model class. The exact definition of the stochastic complexity has gone through several evolutionary steps. The latest instantation is based on the so-called Normalized Maximum Likelihood (NML) distribution which has been shown to possess several important theoretical properties. However, the applications of this modern version of the MDL have been quite rare because of computational complexity problems, i.e., for discrete data, the definition of NML involves an exponential sum, and in the case of continuous data, a multi-dimensional integral usually infeasible to evaluate or even approximate accurately. In this doctoral dissertation, we present mathematical techniques for computing NML efficiently for some model families involving discrete data. We also show how these techniques can be used to apply MDL in two practical applications: histogram density estimation and clustering of multi-dimensional data.

Modeling ecological and evolutionary processes in spatially structured populations

Many species inhabit fragmented landscapes, resulting either from anthropogenic or from natural processes. The ecological and evolutionary dynamics of spatially structured populations are affected by a complex interplay between endogenous and exogenous factors. The metapopulation approach, simplifying the landscape to a discrete set of patches of breeding habitat surrounded by unsuitable matrix, has become a widely applied paradigm for the study of species inhabiting highly fragmented landscapes. In this thesis, I focus on the construction of biologically realistic models and their parameterization with empirical data, with the general objective of understanding how the interactions between individuals and their spatially structured environment affect ecological and evolutionary processes in fragmented landscapes. I study two hierarchically structured model systems, which are the Glanville fritillary butterfly in the Åland Islands, and a system of two interacting aphid species in the Tvärminne archipelago, both being located in South-Western Finland. The interesting and challenging feature of both study systems is that the population dynamics occur over multiple spatial scales that are linked by various processes. My main emphasis is in the development of mathematical and statistical methodologies. For the Glanville fritillary case study, I first build a Bayesian framework for the estimation of death rates and capture probabilities from mark-recapture data, with the novelty of accounting for variation among individuals in capture probabilities and survival. I then characterize the dispersal phase of the butterflies by deriving a mathematical approximation of a diffusion-based movement model applied to a network of patches. I use the movement model as a building block to construct an individual-based evolutionary model for the Glanville fritillary butterfly metapopulation. I parameterize the evolutionary model using a pattern-oriented approach, and use it to study how the landscape structure affects the evolution of dispersal. For the aphid case study, I develop a Bayesian model of hierarchical multi-scale metapopulation dynamics, where the observed extinction and colonization rates are decomposed into intrinsic rates operating specifically at each spatial scale. In summary, I show how analytical approaches, hierarchical Bayesian methods and individual-based simulations can be used individually or in combination to tackle complex problems from many different viewpoints. In particular, hierarchical Bayesian methods provide a useful tool for decomposing ecological complexity into more tractable components.

Theoretical modeling of ionospheric electrodynamics including induction effects

This thesis deals with theoretical modeling of the electrodynamics of auroral ionospheres. In the five research articles forming the main part of the thesis we have concentrated on two main themes: Development of new data-analysis techniques and study of inductive phenomena in the ionospheric electrodynamics. The introductory part of the thesis provides a background for these new results and places them in the wider context of ionospheric research. In this thesis we have developed a new tool (called 1D SECS) for analysing ground based magnetic measurements from a 1-dimensional magnetometer chain (usually aligned in the North-South direction) and a new method for obtaining ionospheric electric field from combined ground based magnetic measurements and estimated ionospheric electric conductance. Both these methods are based on earlier work, but contain important new features: 1D SECS respects the spherical geometry of large scale ionospheric electrojet systems and due to an innovative way of implementing boundary conditions the new method for obtaining electric fields can be applied also at local scale studies. These new calculation methods have been tested using both simulated and real data. The tests indicate that the new methods are more reliable than the previous techniques. Inductive phenomena are intimately related to temporal changes in electric currents. As the large scale ionospheric current systems change relatively slowly, in time scales of several minutes or hours, inductive effects are usually assumed to be negligible. However, during the past ten years, it has been realised that induction can play an important part in some ionospheric phenomena. In this thesis we have studied the role of inductive electric fields and currents in ionospheric electrodynamics. We have formulated the induction problem so that only ionospheric electric parameters are used in the calculations. This is in contrast to previous studies, which require knowledge of the magnetospheric-ionosphere coupling. We have applied our technique to several realistic models of typical auroral phenomena. The results indicate that inductive electric fields and currents are locally important during the most dynamical phenomena (like the westward travelling surge, WTS). In these situations induction may locally contribute up to 20-30% of the total ionospheric electric field and currents. Inductive phenomena do also change the field-aligned currents flowing between the ionosphere and magnetosphere, thus modifying the coupling between the two regions.

Essays on the Modeling and Prediction of Volatility and Higher Moments of Stock Returns (summary section only)

One of the most fundamental and widely accepted ideas in finance is that investors are compensated through higher returns for taking on non-diversifiable risk. Hence the quantification, modeling and prediction of risk have been, and still are one of the most prolific research areas in financial economics. It was recognized early on that there are predictable patterns in the variance of speculative prices. Later research has shown that there may also be systematic variation in the skewness and kurtosis of financial returns. Lacking in the literature so far, is an out-of-sample forecast evaluation of the potential benefits of these new more complicated models with time-varying higher moments. Such an evaluation is the topic of this dissertation. Essay 1 investigates the forecast performance of the GARCH (1,1) model when estimated with 9 different error distributions on Standard and Poor’s 500 Index Future returns. By utilizing the theory of realized variance to construct an appropriate ex post measure of variance from intra-day data it is shown that allowing for a leptokurtic error distribution leads to significant improvements in variance forecasts compared to using the normal distribution. This result holds for daily, weekly as well as monthly forecast horizons. It is also found that allowing for skewness and time variation in the higher moments of the distribution does not further improve forecasts. In Essay 2, by using 20 years of daily Standard and Poor 500 index returns, it is found that density forecasts are much improved by allowing for constant excess kurtosis but not improved by allowing for skewness. By allowing the kurtosis and skewness to be time varying the density forecasts are not further improved but on the contrary made slightly worse. In Essay 3 a new model incorporating conditional variance, skewness and kurtosis based on the Normal Inverse Gaussian (NIG) distribution is proposed. The new model and two previously used NIG models are evaluated by their Value at Risk (VaR) forecasts on a long series of daily Standard and Poor’s 500 returns. The results show that only the new model produces satisfactory VaR forecasts for both 1% and 5% VaR Taken together the results of the thesis show that kurtosis appears not to exhibit predictable time variation, whereas there is found some predictability in the skewness. However, the dynamic properties of the skewness are not completely captured by any of the models.

Essays on Risk Modeling. Applications to Portfolio and Risk Management (summary section only)

In this thesis we deal with the concept of risk. The objective is to bring together and conclude on some normative information regarding quantitative portfolio management and risk assessment. The first essay concentrates on return dependency. We propose an algorithm for classifying markets into rising and falling. Given the algorithm, we derive a statistic: the Trend Switch Probability, for detection of long-term return dependency in the first moment. The empirical results suggest that the Trend Switch Probability is robust over various volatility specifications. The serial dependency in bear and bull markets behaves however differently. It is strongly positive in rising market whereas in bear markets it is closer to a random walk. Realized volatility, a technique for estimating volatility from high frequency data, is investigated in essays two and three. In the second essay we find, when measuring realized variance on a set of German stocks, that the second moment dependency structure is highly unstable and changes randomly. Results also suggest that volatility is non-stationary from time to time. In the third essay we examine the impact from market microstructure on the error between estimated realized volatility and the volatility of the underlying process. With simulation-based techniques we show that autocorrelation in returns leads to biased variance estimates and that lower sampling frequency and non-constant volatility increases the error variation between the estimated variance and the variance of the underlying process. From these essays we can conclude that volatility is not easily estimated, even from high frequency data. It is neither very well behaved in terms of stability nor dependency over time. Based on these observations, we would recommend the use of simple, transparent methods that are likely to be more robust over differing volatility regimes than models with a complex parameter universe. In analyzing long-term return dependency in the first moment we find that the Trend Switch Probability is a robust estimator. This is an interesting area for further research, with important implications for active asset allocation.

Modeling Nonlinearities and Asymmetries in Asset Pricing

Financial time series tend to behave in a manner that is not directly drawn from a normal distribution. Asymmetries and nonlinearities are usually seen and these characteristics need to be taken into account. To make forecasts and predictions of future return and risk is rather complicated. The existing models for predicting risk are of help to a certain degree, but the complexity in financial time series data makes it difficult. The introduction of nonlinearities and asymmetries for the purpose of better models and forecasts regarding both mean and variance is supported by the essays in this dissertation. Linear and nonlinear models are consequently introduced in this dissertation. The advantages of nonlinear models are that they can take into account asymmetries. Asymmetric patterns usually mean that large negative returns appear more often than positive returns of the same magnitude. This goes hand in hand with the fact that negative returns are associated with higher risk than in the case where positive returns of the same magnitude are observed. The reason why these models are of high importance lies in the ability to make the best possible estimations and predictions of future returns and for predicting risk.