The aim of this thesis is to explore clustering algorithms of machine unsupervised learning, which can be used for image database classification by similarity. For chosen clustering algorithms is written up a theoretical basis. For better classification of used database this thesis deals with different methods of image preprocessing. With these methods the features from image are extracted. Next the thesis solves of implementation of preprocessing methods and practical application of clustering algorithms. In practical part is programmed aplication in Python programming language, which classifies the database of images into classes by similarity. The thesis tests all of used methods and at the end of the thesis is processed searches of results.

This thesis deals with the field of Recommendation systems using Deep Neural Networks and their use in book recommendation. There are the main traditional recommender systems analysed and their representations are summarized, as well as systems with more advancec techniques based on machine learning.. The core of the thesis is the use of convolutional neural networks for natural language processing and the creation of a book recommendation system. Suggested system make recommendation based on user data, including user reviews and book data, including full texts.

This thesis deals with discrete wavelet transform of a two dimensional discrete signal using the CDF9/7 wavelet. It contains theoretical principles of discrete wavelet transformation, a description of the convolution and Lifting calculation principle, characteristics of the CDF9/7 wavelet, the possibilities of processing a two dimensional input signal and boundary handling. Both principles (convolution, Lifting) of the calculation of two dimensional discrete wavelet transformation have been implemented within the thesis using the CDF9/7 wavelet in the C++ and Java programming languages. Four applications have been created that way which have been described in detail and then compared with one another especially in light of the calculation speed of discrete wavelet transformation. Finally an analysis of the acquired results and an evaluation of the whole thesis has been performed.

This work deals with a computer simulation of an acoustic room using its impulse response. Two different approaches to the simulation are described with their pros and cons and then the work is focused on the physical approach, which uses room’s impulse response during the simulation. Several methods for the extraction of the impulse response of the acoustic room are mentioned with their conditions of use. The detailed description of various algorithms for a real time convolution computing is followed by the cost analysis of frequency domain block convolution algorithms. Several algorithms are chosen, implemented and tested in Matlab environment. Then the most effective of them is chosen to be implemented in VST technology as the plug in module for real time room simulation.

This thesis deals with a simulation of analogue audio effects using the nonlinear models that replace the analogue nonlinear devices in discrete domain. The thesis describes Volterra system model and simplified Volterra system model that can be realized in two ways, either Wiener model, or Hammerstein model. The method for the analysis and modeling of audio and acoustic nonlinear systems is presented in this thesis. This method allows through knowledge of the input swept-sine signal and the response of the analogue nonlinear system to the input signal to determine the coefficients of the discrete nonlinear system. This allows simulating the analogue nonlinear system in discrete domain. The method was first tested and then used successfully for simulation of the analogue nonlinear system in discrete domain. Concretely, it was simulated a musical guitar effect of the type of distortion. Last part of this thesis is devoted a description of VST technology and an implementation of VST plug-in module, which realizations Hammerstein model.

This thesis deals with image object detection and its classification into classes. Classification is provided by models of framework for deep learning BVLC/Caffe. Object detection is provided by AlpacaDB/selectivesearch and belltailjp/selective_search_py algorithms. One of results of this thesis is modification and usage of deep convolutional neural network AlexNet in BVLC/Caffe framework. This model was trained with precision 51,75% for classification into 1 000 classes. Then it was modified and trained for classification into 20 classes with precision 75.50%. Contribution of this thesis is implementation of graphical interface for object detction and their classification into classes, which is implemented as aplication based on web server in Python language. Aplication integrates object detection algorithms mentioned abowe with classification with help of BVLC/Caffe. Resulting aplication can be used for both object detection (and classification) and for fast verification of any classification model of BVLC/Caffe. This aplication was published on server GitHub under license Apache 2.0 so it can be further implemented and used.

In this paper we deal with the `strength' of ill-posedness for ill-posed linear operator equations Ax = y in Hilbert spaces, where we distinguish according_to_M. Z. Nashed [15] the ill-posedness of type I if A is not compact, but we have R(A) 6= R(A) for the range R(A) of A; and the ill-posedness of type II for compact operators A: From our considerations it seems to follow that the problems with noncompact operators A are not in general `less' ill-posed than the problems with compact operators. We motivate this statement by comparing the approximation and stability behaviour of discrete least-squares solutions and the growth rate of Galerkin matrices in both cases. Ill-posedness measures for compact operators A as discussed in [10] are derived from the decay rate of the nonincreasing sequence of singular values of A. Since singular values do not exist for noncompact operators A; we introduce stability rates in order to have a common measure for the compact and noncompact cases. Properties of these rates are illustrated by means of convolution equations in the compact case and by means of equations with multiplication operators in the noncompact case. Moreover using increasing rearrangements of the multiplier functions specific measures of ill-posedness called ill-posedness rates are considered for the multiplication operators. In this context, the character of sufficient conditions providing convergence rates of Tikhonov regularization are compared for compact operators and multiplication operators.

<p>U ovoj tezi je definisana uop&scaron;tena konvolucija koja pripada domenu pseudo-analize i ima veliku primenu u mnogim matematičkim teorijama, npr. u proba-bilističkim metričkim prostorima, PDJ, teorijama odlučivanja, sistema, kontrole i fazi brojeva. Dokazane su bitne osobine ove operacije sa funkcijama. Dokazana je veza izmedju pseudo-konvolucija baziranih na poluprstenima različitih klasa Definisana je (5, C/)-konvolucija bazirana na uslovno distributivnom poluprstenu ([0,1], S, U)).Dat je jo&scaron; jedan vid uop&scaron;tenja konvolucije baziran na uop&scaron;tenim pseudo-operacijama.</p> / <p>In this thesis the generalized convolution have been defined. This operation with functions has applications in different mathematical theo&shy; ries, for example in Probabilistic Metric Spaces, PDE, System and Control Theory, Fuzzy numbers. Some basic properties of this operation has been proved, as well as connection between generalized convolutions based on dif&shy;ferent classes of semirings. (5, U)-convolution has been defined, as well as convolution based on generalized pseudo-operations.</p>

The IEEE 802.11a protocol is part of the IEEE 802 protocols for implementing WLAN Wi- Fi computer communications in various frequencies. These protocols find applications worldwide, covering a wide range of devices like mobile phones, computers, laptops, household appliances, etc. Since wireless communication is being used, data that is transmitted is susceptible to noise. As a means to recover from noise, the data transmitted is encoded using convolutional encoding and correspondingly decoded on the receiver side. The decoder used is the Viterbi decoder, in the PHY layer of the protocol. This thesis investigates soft-decision Viterbi decoder implementations that meet the requirements of the IEEE 802.11a protocol. It aims to implement a rate-flexible design as a coarse grain re-configurable architecture using the SiLago framework. SiLago is a modular approach towards ASIC design. Components are designed as hardened blocks, which means they are synthesised and pre-verified. Each block is also abuttable like LEGO blocks, which allows users to connect compatible blocks and make designs specific to their requirements, while getting performance similar to that of traditional ASICs. This approach significantly reduces the design costs, as verification is a one-time task. The thesis discusses the strongly connected trellis Viterbi decoding algorithm and proposes a design for a soft decision Viterbi decoder. The proposed design meets the throughput requirements of the communication protocol and it can be reconfigured to work for 45 different code rates, with programmable soft decision width and parallelism. The algorithm used is compared against MATLAB for its BER performance. Results from RTL simulations, advantages and disadvantages of the proposed design are discussed. Recommendations for future improvements are also made. / IEEE 802.11a-protokollet är en del av IEEE 802-protokollen för att implementera WLAN Wi-Fi-datorkommunikation i olika frekvenser. Dessa protokoll används i applikationer över hela världen som täcker ett brett spektrum av produkter som mobiltelefoner, datorer, bärbara datorer, hushållsapparater etc. Eftersom trådlös kommunikation används är data som överförs känslig för brus. Som ett medel för att återhämta sig från brus kodas överförd data med hjälp av faltningskodning och avkodas på motsvarande sätt på mottagarsidan. Den avkodare som används är Viterbi-avkodaren, i PHY-skiktet i protokollet. Denna avhandling undersöker mjuka beslut Viterbi avkodarimplementeringar som uppfyller kraven i IEEE 802.11a protokollet. Det syftar till att implementera en hastighetsflexibel design som en grovkornig konfigurerbar arkitektur som använder SiLago ramverket. SiLago är ett modulärt synsätt på ASIC design. Komponenterna är utformade som härda block, vilket innebär att de är syntetiserade och förverifierade. Varje block kan också kopplas ihop, som LEGO block, vilket gör det möjligt för användare att ansluta kompatibla block och göra designer som är specifika för deras krav, samtidigt som de får prestanda som liknar traditionella ASICs. Detta tillvägagångssätt minskar designkostnaderna avsevärt, eftersom verifiering är en engångsuppgift. Avhandlingen diskuterar den starkt kopplade trellis Viterbi-avkodningsalgoritmen och föreslår en design för en mjuk Viterbi-avkodare. Den föreslagna designen uppfyller kommunikationsprotokollets genomströmningskrav och den kan konfigureras om för att fungera för 45 olika kodhastigheter, med programmerbar mjuk beslutsbredd och parallellitet. Algoritmen som används jämförs mot MATLAB för dess BER-prestanda. Resultat från RTL-simuleringar, fördelar och nackdelar med den föreslagna designen diskuteras. Rekommendationer för framtida förbättringar görs också.

Depth of field is a naturally occurring effect in lenses describing the distance between theclosest and furthest object that appears in focus. The effect is commonly used in film andphotography to direct a viewers focus, give a scene more complexity, or to improve aes-thetics. In computer graphics, the same effect is possible, but since there are no naturaloccurrences of lenses in the virtual world, other ways are needed to achieve it. There aremany different approaches to simulate depth of field, but not all are suited for real-time usein computer games. In this thesis, multiple methods are explored and compared to achievedepth of field in real-time with a focus on computer games. The aspect of bokeh is alsocrucial when considering depth of field, so during the thesis, a method to simulate a bokeheffect similar to reality is explored. Three different methods based on the same approachwas implemented to research this subject, and their time and memory complexity weremeasured. A questionnaire was performed to measure the quality of the different meth-ods. The result is three similar methods, but with noticeable differences in both quality andperformance. The results give the reader an overview of different methods and directionsfor implementing it on their own, based on which requirements suits them.