Class imbalanced datasets constitute a significant portion of the machine learning problems of interest, where recog­nizing the ‘rare class’ is the primary objective for most applications. Traditional linear machine learning algorithms are often not effective in recognizing the rare class. In this research work, a specifically optimized feed-forward artificial neural network (ANN) is proposed and developed to train from moderate to highly imbalanced datasets. The proposed methodology deals with the difficulty in classification task in multiple stages—by optimizing the training dataset, modifying kernel function to generate the gram matrix and optimizing the NN structure. First, the training dataset is extracted from the available sample set through an iterative process of selective under-sampling. Then, the proposed artificial NN comprises of a kernel function optimizer to specifically enhance class boundaries for imbalanced datasets by conformally transforming the kernel functions. Finally, a single hidden layer weighted neural network structure is proposed to train models from the imbalanced dataset. The proposed NN architecture is derived to effectively classify any binary dataset with even very high imbalance ratio with appropriate parameter tuning and sufficient number of processing elements. Effectiveness of the proposed method is tested on accuracy based performance metrics, achieving close to and above 90%, with several imbalanced datasets of generic nature and compared with state of the art methods. The proposed model is also used for classification of a 25GB computed tomographic colonography database to test its applicability for big data. Also the effectiveness of under-sampling, kernel optimization for training of the NN model from the modified kernel gram matrix representing the imbalanced data distribution is analyzed experimentally. Computation time analysis shows the feasibility of the system for practical purposes. This report is concluded with discussion of prospect of the developed model and suggestion for further development works in this direction.

This PhD thesis aims at exploring price convergence in the European Union with a particular emphasis paid to the Czech Republic and new EU member states. Fundamental issues are discussed in the first chapter, starting with the notion and term `convergence' since many alternative definitions have been proposed in the literature. Apart from that, main indicators utilized when investigating price convergence are defined (for example purchasing power parity/purchasing power standard, PPP/PPS, comparative price level, CPL) and a brief review of the literature is added. The second chapter deals with several issues accompanying price convergence in general and in transforming countries in particular such as the club convergence hypothesis, issues of tradability, availability of datasets and their strenghts and weaknesses, the link between price levels and rates of inflation, and determinants. Both `standard' and `modern' approaches are utilized in the last chapter so that several hypotheses can be verified. For the sake of comparability, individual CPLs for EU-27 countries for the period 1995(9)-2011 are employed. Firstly, stylised facts for both old EU and NMS are presented (including effects stemming from the on-going financial crisis). Secondly, the club convergence hypothesis is examined with help of two different ways - cluster analysis and the Phillips-Sul test (both for the EU and its `subgroups'). Both of them do confirm the existence of convergence clubs in the EU (including its old and new part). Following the previous findings, a somewhat broader and richer view on price level dynamics is supplemented via utilization of the so-called Stochastic kernel (Quah, 1993). This methodology shows both convergence and divergence (divergence/polarization/stratification) in the EU. Finally, the last section of this chapter is focused on a thorough search for determinants of price levels in the EU. The Bayesian approach is employed (Bayesian model averaging, BMA) and our results confirm both the importance of both `traditional' determinants such as labour costs and output gap and new ones such as broadly defined institutional factors. Main findings of this thesis are summarized and commented in the conclusion aiming at providing implications for policymakers and some guidance for future research.