Title: Machine Learning for Credit Scoring Author: Elena Myazina Department / Institute: Department of Theoretical Computer Science and Mathematical Logic Supervisor of the master thesis: Mgr. Martin Pilát, Ph.D, Department of Theoretical Computer Science and Mathematical Logic Abstract: Credit scoring is a technique used by banks to evaluate their clients who ask for different types of loan. Its goal is to predict, whether a given client will pay their loan or not. Traditionally, mathematical models based on logistic regression are used for this task. In this thesis, we approach the problem of credit scoring from a machine learning point of view. We investigate several machine learning methods (including neural networks, random forests, support vector machines and other), and evaluate their performance for the credit scoring task on three publicly available datasets.. Keywords: machine learning, credit scoring,logistic regression, neural networks, random forest

Recently, machine learning has been put into connection with a field called ,,Big Data'' more and more. Usually, in this field, a lot of data is available and we need to gather useful information based on this data. Nowadays, when still more and more data is generated by use of mobile phones, credit cards, etc., a need for high-performance methods is serious. In this work, we describe six different methods that serve this purpose. These are logistic regression, neural networks and deep neural networks, bagging, boosting and stacking. Last three methods compose a group called Ensemble Learning. We apply all six methods on real data, which were generously provided by one of the loan providers. These methods can help them to distinguish between good and bad potential takers of loans, when the decision about the loan is being made. Lastly, the results of particular methods are compared and we also briefly outline possible ways of interpretation.

This thesis focuses on implementation of an algorithm for retinal vessel segmentation in high resolution retinal images.A neural network with two hidden layers was used as the method. A total of 7 features were obtained from matched filtering based on vessel thickness, texture analysis and individual pixels brightness. Within the thesis, the whole database was manually annotated for the implementation of the algorithm and the results. The achieved mean sensitivity reached 80%, specificity 70% and Dice coefficient is 59%.

The goal of this study is to learn about methods in object classification in medicine and find out what are these methods about. Focusing on functionality and reliability of these methods whith datafile from the medicine compartment after making the algorithm in MATLAB. In form of siple tests, put the touch everyone of classification procedure and find out in which they excel and in which they lags. The choice of input data parametres is very important, this will be tested and noted in conclusions.

This master thesis is written in collaboration with NIC.CZ company. It describes basic principles of DNS system and properties of DNS traffic. It's goal an implementation of DNS anomaly classifier and its evaluation in practice.

The goal of this work was to design fast and simple methods for processing point-cloud-data of urban areas for virtual reality applications. For the visualization of methods, we developed a simple renderer written in C++ and HLSL. The renderer is based on DirectX 11. For point-cloud processing, we designed a method based on height-histograms for filtering ground points out of point cloud. We also proposed a parallel method for point cloud segmentation based on the region growing algorithm. The individual segments are then tested by simple rules to check if it is or it is not corresponding to a predefined object.

7 Automatic extraction of buildings and imprevious areas from very high resolution data in suburban area of Prague Abstract Nowdays, when the very high resolution satellite imagery and airbone laser scanning data have became more accessible, the possibility of their use for different types of applications increased also. With a rapid development of urban hinterlands the demand to monitor these areas increases also with the goal to avoid uncoordinated construction. This work therefore focuses on an object oriented based classification in order to design its own methodological approach for the extraction of buildings and imprevious areas in selected areas of commercial suburbanization in the Prague hinterland. The aim is among others to find the classification rules for distinguishing different types of roofs and impervious areas depending on the material, shape characteristic etc. The literature overview summarizes methods of buildings and imprevious areas classification and extraction using very high resolution optical data and elevation data. Very high resolution QuickBird imagery and airbone laser scanning LIDAR elevation data and object oriented classification methods were used for and analysis of selected commercial suburbanization model areas in Prague hinterland. The proposed methodology uses...

Title: Validity assessment of the grading scale evaluation in a new bouldering area. Objectives: The aim of this work is to assess the grading scale validity of bouldering routes in a new area. Methods: Routes grading was verified by a group of six men and four women. The evaluation took place during one continuous visit in the rocks with an individual rest period between problems. The classification of each of the 10 ascents was conducted immediately after finishing the specific problem. It was done individually, without the possibility to consult their guesses and ideas. The proposed difficulties of the classified routes ranged from 12 to 17 IRCRA. Results: The average rating of the route difficulty classification was 0.89 IRCRA points higher than the initial suggestion, with minor differences between men and women. The influence of climbers' performance on their ability to assess routes' difficulty was statistically insignificant. Conclusion: The first climbers of boulders in the area of Goethe rock underestimate the climbing grades, which proves that the determination of difficulty in local climbing centers can be significantly skewed compared to established standards. Keywords: bouldering, grading scales, IRCRA, validity

Plastic filaments are used in a fastly growing industry of 3D printing using the Fused Filament Fabrication (FFF) method. A poor quality of spool winding can negatively impact the printing process. On the other hand, producing high-quality filament winding is surprisingly difficult to achieve consistently. The thesis proposes a holistic approach to inspect winding quality during the winding process. We suggest tracking reflections of bright visible light. This method seems robust enough to track filament color from black to white and even transparent materials. Furthermore, it is possible to run everything on cheap and widely available Raspberry Pi 4 B with Camera Module v2. The system uses classical computer vision approaches for filtering, segmentation, and inter-frame tracking of individual filament strands between video frames. It was confirmed to be fast enough to process 30 FPS footage directly on the Raspberry Pi in real-time. Additionally, the GUI tool for quick dataset annotation of spool winding images was created along with a small dataset. Both might be useful for the future development of a system, which would predict the quality issues earlier when corrective action can still be carried out to prevent them. 1

This diploma thesis deals with the issue of evaluation and classification of pupils in art education. The aim of the diploma thesis is to describe the form of evaluation and classification of art education at the second stage of primary school for selected teachers. Following the aim of the work, the following questions were asked: How do teachers evaluate pupils in art education? Does their evaluation subsequently correspond to the classification? According to art teachers, is classification necessary? What do teachers emphasize when evaluating a student's artistic performance? Within this work, qualitative research was performed. Selected research methods include observation of lessons of selected art teachers and subsequent interviews with them. Each of the teachers teaches art education and another subject at primary school. The main conclusions of the research include the finding that according to the respondents, evaluation is an activity that helps teachers in critical evaluation of student work, evaluation criteria and form of evaluation are not uniform among respondents and ideas about evaluation and classification often differ from the observed reality. Furthermore, respondents evaluate and classify in different ways. However, they clearly agreed on the importance of classification in art...