Developments of the generative topographic mapping

The generative topographic mapping (GTM) model was introduced by Bishop et al. (1998, Neural Comput. 10(1), 215-234) as a probabilistic re- formulation of the self-organizing map (SOM). It offers a number of advantages compared with the standard SOM, and has already been used in a variety of applications. In this paper we report on several extensions of the GTM, including an incremental version of the EM algorithm for estimating the model parameters, the use of local subspace models, extensions to mixed discrete and continuous data, semi-linear models which permit the use of high-dimensional manifolds whilst avoiding computational intractability, Bayesian inference applied to hyper-parameters, and an alternative framework for the GTM based on Gaussian processes. All of these developments directly exploit the probabilistic structure of the GTM, thereby allowing the underlying modelling assumptions to be made explicit. They also highlight the advantages of adopting a consistent probabilistic framework for the formulation of pattern recognition algorithms.

Gaussian processes:iterative sparse approximations

In recent years there has been an increased interest in applying non-parametric methods to real-world problems. Significant research has been devoted to Gaussian processes (GPs) due to their increased flexibility when compared with parametric models. These methods use Bayesian learning, which generally leads to analytically intractable posteriors. This thesis proposes a two-step solution to construct a probabilistic approximation to the posterior. In the first step we adapt the Bayesian online learning to GPs: the final approximation to the posterior is the result of propagating the first and second moments of intermediate posteriors obtained by combining a new example with the previous approximation. The propagation of em functional forms is solved by showing the existence of a parametrisation to posterior moments that uses combinations of the kernel function at the training points, transforming the Bayesian online learning of functions into a parametric formulation. The drawback is the prohibitive quadratic scaling of the number of parameters with the size of the data, making the method inapplicable to large datasets. The second step solves the problem of the exploding parameter size and makes GPs applicable to arbitrarily large datasets. The approximation is based on a measure of distance between two GPs, the KL-divergence between GPs. This second approximation is with a constrained GP in which only a small subset of the whole training dataset is used to represent the GP. This subset is called the em Basis Vector, or BV set and the resulting GP is a sparse approximation to the true posterior. As this sparsity is based on the KL-minimisation, it is probabilistic and independent of the way the posterior approximation from the first step is obtained. We combine the sparse approximation with an extension to the Bayesian online algorithm that allows multiple iterations for each input and thus approximating a batch solution. The resulting sparse learning algorithm is a generic one: for different problems we only change the likelihood. The algorithm is applied to a variety of problems and we examine its performance both on more classical regression and classification tasks and to the data-assimilation and a simple density estimation problems.

Analysis of productivity differences among foreign and domestic firms:evidence from Germany

This paper presents differences in firm-level total factor productivity (TFP) across 22 manufacturing and 17 service industries in Germany over the period 1995–2004. It is an attempt to study whether and to what extent foreign multinational enterprises (MNEs) are more productive relative to German firms. As well as distinguishing between foreign and domestic firms, we also distinguish between German MNEs and domestic firms that do not have any foreign presence. Controlling for endogeneity through semi-parametric techniques, our findings indicate considerable heterogeneity in firm performance across types of firms. The foreign/domestic distinction is not as clear cut as has been suggested elsewhere; multinationality is important in explaining productivity differences rather than foreignness.

Exchanging uncertainty: interoperable geostatistics?

Traditionally, geostatistical algorithms are contained within specialist GIS and spatial statistics software. Such packages are often expensive, with relatively complex user interfaces and steep learning curves, and cannot be easily integrated into more complex process chains. In contrast, Service Oriented Architectures (SOAs) promote interoperability and loose coupling within distributed systems, typically using XML (eXtensible Markup Language) and Web services. Web services provide a mechanism for a user to discover and consume a particular process, often as part of a larger process chain, with minimal knowledge of how it works. Wrapping current geostatistical algorithms with a Web service layer would thus increase their accessibility, but raises several complex issues. This paper discusses a solution to providing interoperable, automatic geostatistical processing through the use of Web services, developed in the INTAMAP project (INTeroperability and Automated MAPping). The project builds upon Open Geospatial Consortium standards for describing observations, typically used within sensor webs, and employs Geography Markup Language (GML) to describe the spatial aspect of the problem domain. Thus the interpolation service is extremely flexible, being able to support a range of observation types, and can cope with issues such as change of support and differing error characteristics of sensors (by utilising descriptions of the observation process provided by SensorML). XML is accepted as the de facto standard for describing Web services, due to its expressive capabilities which allow automatic discovery and consumption by ‘naive’ users. Any XML schema employed must therefore be capable of describing every aspect of a service and its processes. However, no schema currently exists that can define the complex uncertainties and modelling choices that are often present within geostatistical analysis. We show a solution to this problem, developing a family of XML schemata to enable the description of a full range of uncertainty types. These types will range from simple statistics, such as the kriging mean and variances, through to a range of probability distributions and non-parametric models, such as realisations from a conditional simulation. By employing these schemata within a Web Processing Service (WPS) we show a prototype moving towards a truly interoperable geostatistical software architecture.

Modelling nonlinear stochastic dynamics in financial time series

For analysing financial time series two main opposing viewpoints exist, either capital markets are completely stochastic and therefore prices follow a random walk, or they are deterministic and consequently predictable. For each of these views a great variety of tools exist with which it can be tried to confirm the hypotheses. Unfortunately, these methods are not well suited for dealing with data characterised in part by both paradigms. This thesis investigates these two approaches in order to model the behaviour of financial time series. In the deterministic framework methods are used to characterise the dimensionality of embedded financial data. The stochastic approach includes here an estimation of the unconditioned and conditional return distributions using parametric, non- and semi-parametric density estimation techniques. Finally, it will be shown how elements from these two approaches could be combined to achieve a more realistic model for financial time series.

Semiparametric smooth-coefficient stoachastic frontier model

This paper proposes a semiparametric smooth-coefficient stochastic production frontier model where all the coefficients are expressed as some unknown functions of environmental factors. The inefficiency term is multiplicatively decomposed into a scaling function of the environmental factors and a standard truncated normal random variable. A testing procedure is suggested for the relevance of the environmental factors. Monte Carlo study shows plausible ¯nite sample behavior of our proposed estimation and inference procedure. An empirical example is given, where both the semiparametric and standard parametric models are estimated and results are compared.

Evaluating the environmental Kuznets curve

We examine the empirical evidence for an environmental Kuznets curve using a semiparametric smooth coefficient regression model that allows us to incorporate flexibility in the parameter estimates, while maintaining the basic econometric structure that is typically used to estimate the pollution-income relationship. This allows us to assess the sensitivity to parameter heterogeneity of typical parametric models used to estimate the relationship between pollution and income, as well as identify why the results from such models are seldom found to be robust. Our results confirm that the resulting relationship between pollution and income is fragile; we show that the estimated pollution-income relationship depends substantially on the heterogeneity of the slope coefficients and the parameter values at which the relationship is evaluated. Different sets of parameters obtained from the semiparametric model give rise to many different shapes for the pollution-income relationship that are commonly found in the literature.

Innovation and the survival of new firms across British regions

This paper analyses the survival of the complete cohort of more than 162,000 limited companies incorporated in Britain in 2001 over the subsequent five-year period. For this purpose, we estimate firms' hazards of failure and survival functions using nonparametric and semi-parametric techniques. The paper focuses on two important policy-related issues.The first is to what extent survival rates vary across regions in Britain. A second, and related, policy issue concerns innovation. The data available allows us to look at the intellectual property (IP) activity of all British firms, including that of the 162,000 new firms in 2001. The results indicate substantial differences in survival rates across regions, and also that IP activity is associated with a higher probability of survival. These differences across regions, and the importance of IP activity, remain when we condition on a large range of regional, industry and firm-level characteristics shifting firms' hazards of failure.

Analytical and numerical analysis of 40 Gbit/s regenerated transmissions using simple black box optical regenerator

A simplified (without phase modulator) scheme of a black box optical regenerator is proposed, where an appropriate nonlinear propagation is used to enhance regeneration. Applying semi-theoretical models the authors optimise and demonstrate feasibility of error-free long distance transmission at 40 Gbit/s.

Three essays on a longitudinal analysis of business start-ups using the Kauffman Firm Survey

This dissertation focused on the longitudinal analysis of business start-ups using three waves of data from the Kauffman Firm Survey. ^ The first essay used the data from years 2004-2008, and examined the simultaneous relationship between a firm's capital structure, human resource policies, and its impact on the level of innovation. The firm leverage was calculated as, debt divided by total financial resources. Index of employee well-being was determined by a set of nine dichotomous questions asked in the survey. A negative binomial fixed effects model was used to analyze the effect of employee well-being and leverage on the count data of patents and copyrights, which were used as a proxy for innovation. The paper demonstrated that employee well-being positively affects the firm's innovation, while a higher leverage ratio had a negative impact on the innovation. No significant relation was found between leverage and employee well-being.^ The second essay used the data from years 2004-2009, and inquired whether a higher entrepreneurial speed of learning is desirable, and whether there is a linkage between the speed of learning and growth rate of the firm. The change in the speed of learning was measured using a pooled OLS estimator in repeated cross-sections. There was evidence of a declining speed of learning over time, and it was concluded that a higher speed of learning is not necessarily a good thing, because speed of learning is contingent on the entrepreneur's initial knowledge, and the precision of the signals he receives from the market. Also, there was no reason to expect speed of learning to be related to the growth of the firm in one direction over another.^ The third essay used the data from years 2004-2010, and determined the timing of diversification activities by the business start-ups. It captured when a start-up diversified for the first time, and explored the association between an early diversification strategy adopted by a firm, and its survival rate. A semi-parametric Cox proportional hazard model was used to examine the survival pattern. The results demonstrated that firms diversifying at an early stage in their lives show a higher survival rate; however, this effect fades over time.^

The Most Complete Photometric Analysis of 548 CALIFA Galaxies

We present an extensive photometric catalog for 548 CALIFA galaxies observed as of the summer of 2015. CALIFA is currently lacking photometry matching the scale and diversity of its spectroscopy; this work is intended to meet all photometric needs for CALIFA galaxies while also identifying best photometric practices for upcoming integral field spectroscopy surveys such as SAMI and MaNGA. This catalog comprises gri surface brightness profiles derived from Sloan Digital Sky Survey (SDSS) imaging, a variety of non-parametric quantities extracted from these pro files, and parametric models fitted to the i-band pro files (1D) and original galaxy images (2D). To compliment our photometric analysis, we contrast the relative performance of our 1D and 2D modelling approaches. The ability of each measurement to characterize the global properties of galaxies is quantitatively assessed, in the context of constructing the tightest scaling relations. Where possible, we compare our photometry with existing photometrically or spectroscopically obtained measurements from the literature. Close agreement is found with Walcher et al. (2014), the current source of basic photometry and classifications of CALIFA galaxies, while comparisons with spectroscopically derived quantities reveals the effect of CALIFA's limited field of view compared to broadband imaging surveys such as the SDSS. The colour-magnitude diagram, star formation main sequence, and Tully-Fisher relation of CALIFA galaxies are studied, to give a small example of the investigations possible with this rich catalog. We conclude with a discussion of points of concern for ongoing integral field spectroscopy surveys and directions for future expansion and exploitation of this work.

Do Peers Affect Student Achievement in China's Secondary Schools (Working Paper 35)

Peer effects have figured prominently in debates on school vouchers, desegregation, ability tracking and anti-poverty programs. Compelling evidence of their existence remains scarce for plaguing endogeneity issues such as selection bias and the reflection problem. This paper is among the first to firmly establish the link between peer performance and student achievement, using a unique dataset from China. We find strong evidence that peer effects exist and operate in a positive and nonlinear manner; reducing the variation of peer performance increases achievement; and our semi-parametric estimates clarify the tradeoffs facing policymakers in exploiting positive peers effects to increase future achievement.

Essays in asset price bubbles

This thesis studies the field of asset price bubbles. It is comprised of three independent chapters. Each of these chapters either directly or indirectly analyse the existence or implications of asset price bubbles. The type of bubbles assumed in each of these chapters is consistent with rational expectations. Thus, the kind of price bubbles investigated here are known as rational bubbles in the literature. The following describes the three chapters. Chapter 1: This chapter attempts to explain the recent US housing price bubble by developing a heterogeneous agent endowment economy asset pricing model with risky housing, endogenous collateral and defaults. Investment in housing is subject to an idiosyncratic risk and some mortgages are defaulted in equilibrium. We analytically derive the leverage or the endogenous loan to value ratio. This variable comes from a limited participation constraint in a one period mortgage contract with monitoring costs. Our results show that low values of housing investment risk produces a credit easing effect encouraging excess leverage and generates credit driven rational price bubbles in the housing good. Conversely, high values of housing investment risk produces a credit crunch characterized by tight borrowing constraints, low leverage and low house prices. Furthermore, the leverage ratio was found to be procyclical and the rate of defaults countercyclical consistent with empirical evidence. Chapter 2: It is widely believed that financial assets have considerable persistence and are susceptible to bubbles. However, identification of this persistence and potential bubbles is not straightforward. This chapter tests for price bubbles in the United States housing market accounting for long memory and structural breaks. The intuition is that the presence of long memory negates price bubbles while the presence of breaks could artificially induce bubble behaviour. Hence, we use procedures namely semi-parametric Whittle and parametric ARFIMA procedures that are consistent for a variety of residual biases to estimate the value of the long memory parameter, d, of the log rent-price ratio. We find that the semi-parametric estimation procedures robust to non-normality and heteroskedasticity errors found far more bubble regions than parametric ones. A structural break was identified in the mean and trend of all the series which when accounted for removed bubble behaviour in a number of regions. Importantly, the United States housing market showed evidence for rational bubbles at both the aggregate and regional levels. In the third and final chapter, we attempt to answer the following question: To what extend should individuals participate in the stock market and hold risky assets over their lifecycle? We answer this question by employing a lifecycle consumption-portfolio choice model with housing, labour income and time varying predictable returns where the agents are constrained in the level of their borrowing. We first analytically characterize and then numerically solve for the optimal asset allocation on the risky asset comparing the return predictability case with that of IID returns. We successfully resolve the puzzles and find equity holding and participation rates close to the data. We also find that return predictability substantially alter both the level of risky portfolio allocation and the rate of stock market participation. High factor (dividend-price ratio) realization and high persistence of factor process indicative of stock market bubbles raise the amount of wealth invested in risky assets and the level of stock market participation, respectively. Conversely, rare disasters were found to bring down these rates, the change being severe for investors in the later years of the life-cycle. Furthermore, investors following time varying returns (return predictability) hedged background risks significantly better than the IID ones.

Football Video Annotation Based on Player Motion Recognition Using Enhanced Entropy

This paper presents a semi-parametric Algorithm for parsing football video structures. The approach works on a two interleaved based process that closely collaborate towards a common goal. The core part of the proposed method focus perform a fast automatic football video annotation by looking at the enhance entropy variance within a series of shot frames. The entropy is extracted on the Hue parameter from the HSV color system, not as a global feature but in spatial domain to identify regions within a shot that will characterize a certain activity within the shot period. The second part of the algorithm works towards the identification of dominant color regions that could represent players and playfield for further activity recognition. Experimental Results shows that the proposed football video segmentation algorithm performs with high accuracy.

Spatial and temporal country-wide survey of temephos resistance in Brazilian populations of Aedes aegypti

The organophosphate temephos has been the main insecticide used against larvae of the dengue and yellow fever mosquito ( Aedes aegypti ) in Brazil since the mid-1980s. Reports of resistance date back to 1995; however, no systematic reports of widespread temephos resistance have occurred to date. As resistance investigation is paramount for strategic decision-making by health officials, our objective here was to investigate the spatial and temporal spread of temephos resistance in Ae. aegypti in Brazil for the last 12 years using discriminating temephos concentrations and the bioassay protocols of the World Health Organization. The mortality results obtained were subjected to spatial analysis for distance interpolation using semi-variance models to generate maps that depict the spread of temephos resistance in Brazil since 1999. The problem has been expanding. Since 2002-2003, approximately half the country has exhibited mosquito populations resistant to temephos. The frequency of temephos resistance and, likely, control failures, which start when the insecticide mortality level drops below 80%, has increased even further since 2004. Few parts of Brazil are able to achieve the target 80% efficacy threshold by 2010/2011, resulting in a significant risk of control failure by temephos in most of the country. The widespread resistance to temephos in Brazilian Ae. aegypti populations greatly compromise effective mosquito control efforts using this insecticide and indicates the urgent need to identify alternative insecticides aided by the preventive elimination of potential mosquito breeding sites.