This work deals with image processing and face detection. Includes approaches to the problems of image processing. Furthermore, it focuses mainly on the choice of appropriate libraries and implementation of algorithms able to detect faces from the input image data.

One of new techniques in the field of analytical chemistry, which has more and more practical use, is mass spectrometry imaging. With its ability to record representation of substances in samples during the tissue analyze arise problem with a lot of output data which needs to be handled programmatically. The goal of this work is to create an software for processing and visualization data of new standard imzML. As a part of the work, the field of mass spectrometry, primarily MALDI TOF mass spectrometry, is briefly introduced. There are also introduced some methods for mass spectrometry data preprocessing. The work also contains a summary of current state of available software for processing and visualization of mass spectrometry data. With requests from cooperating laboratory a novel software is designed and implemented, which besides the visualization itself, can preprocess the data for example data smoothing with Savitzky-Golay method, internal calibration or peak detection with continuous wavelet transformation. The software was successfully tested on real data sets.

This thesis deals with text mining. It focuses on problems of document classification and related techniques, mainly data preprocessing. Project also introduces the SVM method, which has been chosen for classification, design and testing of implemented application.

Background: Filtering is a ubiquitous step in the preprocessing of electroencephalographic (EEG) and magnetoencephalographic (MEG) data. Besides the intended effect of the attenuation of signal components considered as noise, filtering can also result in various unintended adverse filter effects (distortions such as smoothing) and filter artifacts. Method: We give some practical guidelines for the evaluation of filter responses (impulse and frequency response) and the selection of filter types (high-pass/low-pass/band-pass/band stop; finite/infinite impulse response, FIR/IIR) and filter parameters (cutoff frequencies, filter order and roll-off, ripple, delay and causality) to optimize signal to-noise ratio and avoid or reduce signal distortions for selected electrophysiological applications. Results: Various filter implementations in common electrophysiology software packages are introduced and discussed. Resulting filter responses are compared and evaluated. Conclusion: We present strategies for recognizing common adverse filter effects and filter artifacts and demonstrate them in practical examples. Best practices and recommendations for the selection and reporting of filter parameters, limitations, and alternatives to filtering are discussed.

Master's thesis focuses on processing fMRI data, which are mapping blood oxygenation level dependence in a state of brain activity. Usable and necessarily preprocessing tech- niques of the data, together with two main analysis approaches are introduced. The area of univariate methods, especially general linear model and multivariate principal or independent component analysis is explained. Practical application of the methods involved on the real fMRI data set is implemented. Relevant results as well as theirs mutual possible comparison is presented.

DNA methylation plays a key role in disease analysis, especially for studies that compare known large scale differences in CpG sites, such as cancer/normal studies or between-tissues studies. However, before any analysis can be done, data normalization and preprocessing of methylation data are required. A useful data preprocessing pipeline for large scale comparisons is Functional Normalization (FunNorm), (Fortin et al., 2014) implemented in the minfi package in R. In FunNorm, the univariate quantiles of the methylated and unmethylated signal values in the raw data are used to preprocess the data. However, although FunNorm has been shown to outperform other preprocessing and data normalization processes for these types of studies, it does not account for the correlation between the methylated and unmethylated signals into account; the focus of this paper is to improve upon FunNorm by taking this correlation into account. The concept of a bivariate quantile is used in this study as an attempt to take the correlation between the methylated and unmethylated signals into consideration. From the bivariate quantiles found, the partial least squares method is then used on these quantiles in this preprocessing. The raw datasets used for this research were collected from the European Molecular Biology Laboratory - European Bioinformatics Institute (EMBL-EBI) website. The results from this preprocessing algorithm were then compared and contrasted to the results from FunNorm. Drawbacks, limitations and future research are then discussed. / Thesis / Master of Science (MSc)

With the development in the area of machine learning, society has become more dependent on applications that build on machine learning techniques. Despite this, there are extensive classification tasks which are still performed by humans. This is time costly and often results in errors. One application in machine learning is text classification which has been researched a lot the past twenty years. Text classification tasks can be automated through the machine learning technique supervised learning which can lead to increased performance compared to manual classification. When handling text data, the data often has to be preprocessed in different ways to assure a good classification. Preprocessing techniques have been shown to increase performance of text classification through supervised learning. Different processing techniques affect the performance differently depending on the choice of learning algorithm and characteristics of the data set.   This thesis investigates how classification accuracy is affected by different learning algorithms and different preprocessing techniques for a specific customer feedback data set. The researched algorithms are Naïve Bayes, Support Vector Machine and Decision Tree. The research is done by experiments with dependency on algorithm and combinations of preprocessing techniques. The results show that spelling correction and removing stop words increase the accuracy for all classifiers while stemming lowers the accuracy for all classifiers. Furthermore, Decision Tree was most positively affected by preprocessing while Support Vector Machine was most negatively affected. A deeper study on why the preprocessing techniques affected the algorithms in such a way is recommended for future work. / I och med utvecklingen inom området maskininlärning har samhället blivit mer beroende av applikationer som bygger på maskininlärningstekniker. Trots detta finns omfattande klassificeringsuppgifter som fortfarande utförs av människor. Detta är tidskrävande och resulterar ofta i olika typer av fel. En  uppgift inom maskininlärning är textklassificering som har forskats mycket i de senaste tjugo åren. Textklassificering kan automatiseras genom övervakad maskininlärningsteknik vilket kan leda till effektiviseringar jämfört med manuell klassificering. Ofta måste textdata förbehandlas på olika sätt för att säkerställa en god klassificering. Förbehandlingstekniker har visat sig öka textklassificeringens prestanda genom övervakad inlärning. Olika förbetningstekniker påverkar prestandan olika beroende på valet av inlärningsalgoritm och egenskaper hos datamängden.  Denna avhandling undersöker hur klassificeringsnoggrannheten påverkas av olika inlärningsalgoritmer och olika förbehandlingstekniker för en specifik datamängd som utgörs av kunddata. De undersökta algoritmerna är naïve Bayes, supportvektormaskin och beslutsträd. Undersökningen görs genom experiment med beroende av algoritm och kombinationer av förbehandlingstekniker. Resultaten visar att stavningskorrektion och borttagning av stoppord ökar noggrannheten för alla klassificerare medan stämming sänker noggrannheten för alla. Decision Tree var dessutom mest positivt påverkad av de olika förbehandlingsmetoderna medan Support Vector Machine påverkades mest negativt. En djupare studie om varför förbehandlingsresultaten påverkat algoritmerna på ett sådant sätt rekommenderas för framtida arbete.

Demand forecasting for sales is a widely researched topic that is essential for a business to prepare for market changes and increase profits. Existing research primarily focus on data that is more suitable for machine learning applications compared to the data accessible to companies lacking prior machine learning experience. This thesis performs demand forecasting on a known sales dataset and a dataset accessed directly from such a company, in the hopes of gaining insights that can help similar companies better utilize machine learning in their business model. LigthGBM, Linear Regression and Random Forest models are used along with several regression error metrics and plots to compare the performance of the two datasets. Both data sets are preprocessed into the same structure based on equivalent features found in each set. The company dataset is determined to be unfit for machine learning forecasting even after preprocessing measures and multiple possible reasons are established. The main contributors are a lack of observations per article and uniformity through the time series.

It is projected that by 2026, most workloads in cloud data centers will be Deep Learning (DL) workloads. However, these workloads pose significant challenges due to their high computational demands, requiring infrastructure and platform advancements to meet DL’s performance, efficiency, and scalability requirements. One emerging problem in large-scale DL is the data stall issue, which occurs when DL models require extensive input data pre-processing, causing CPUs to struggle to keep up with the data consumption demands of GPUs during the training stage. This results in the DL pipeline stalling and GPUs running idle. Our work aims to fundamentally address the data stall issue in modern pre-processing DL pipelines. Traditional solutions involve allocating more CPUs to the pre-processing stage to meet GPU demands, but this approach significantly increases energy con- sumption and provisioning costs. For example, Meta recently disclosed that their DLRM pipeline requires 9 to 55 CPU servers per trainer node, depending on the workload. Our research explores offloading common pre-processing primi- tives to programmable network hardware, specifically Tofino2-equipped switches known for their high bandwidth and energy efficiency, and the Bluefield-2 Smart- NIC. Our initial power measurements demonstrate that Tofino2 and Bluefield-2 achieve 11.6x and 3.0x higher throughput per Watt, respectively, compared to a generic x86 or AMD CPU while performing pre-processing operations. However, due to Tofino2’s limitations in terms of the operations it can perform compared to a CPU, several design optimizations are required to fully exploit the potential of programmable network devices.

No description available.