National Repository of Grey Literature 412 records found  1 - 10nextend  jump to record: Search took 0.00 seconds. 
Online training of deep neural networks for classification
Tumpach, Jiří ; Holeňa, Martin (advisor) ; Kořenek, Jakub (referee)
Deep learning is usually applied to static datasets. If used for classification based on data streams, it is not easy to take into account a non-stationarity. This thesis presents work in progress on a new method for online deep classifi- cation learning in data streams with slow or moderate drift, highly relevant for the application domain of malware detection. The method uses a combination of multilayer perceptron and variational autoencoder to achieve constant mem- ory consumption by encoding past data to a generative model. This can make online learning of neural networks more accessible for independent adaptive sys- tems with limited memory. First results for real-world malware stream data are presented, and they look promising. 1
Evaluation of the land cover in the military training area Libavá using Random Forest classifier
Žďánský, Vít ; Štych, Přemysl (advisor) ; Laštovička, Josef (referee)
Data land cover help us understand nature, how it develops, its uses and the influence that human actions have on it. Thanks to new methods in the remote sensing area, we can record these processes faster and at a larger scale than before. This thesis evaluates accuracy of the Random Forest (RF) and Maximum Likelihood (ML) classifiers using satellite data Sentinel-2 from the military training area Libavá. The military area went through a very specific development and the information regarding natural coverage in the region is missing. The classifier documentation contains 8 classes. The classification results from both algorithms are higher than 80 %. As expected, more accurate results were achieved using the Random Forest classifier. The most accurate classifications were of water surfaces and forests. The least accurate classifications were of agricultural land and sparse vegetation. Other classes varied in accuracy levels. This thesis' results are evaluated using error matrices, overall accuracy and the kappa coefficient. Keywords: classification, Random Forest, Maximum Likelihood, military training area, remote sensing, Sentinel 2, land cover, Libavá
Predictive Modelling with Python
Duda, Jan ; Burgetová, Ivana (referee) ; Zendulka, Jaroslav (advisor)
The main goal of this bachelor thesis is get to know with the data mining and its domain, also with the Knowledge discovery in databases process. It shows the most importnant approaches, which are implemented in Python language afterwards. The case study contains the prediction of index S&P 500 describing stock market developments on the US stock exchange. Both classification and regression models are used for the forecasting. Model evaluation is reached by the Monte Carlo experimental method.
Topic Identification from Spoken TED-Talks
Vašš, Adam ; Ondel, Lucas (referee) ; Kesiraju, Santosh (advisor)
Táto práca sa zaoberá problémom spracovania prirodzeného jazyka a následnej klasifikácie. Použité systémy boli modelované na TED-LIUM korpuse. Systém automatického spracovania jazyka bol modelovaný s použitím sady nástrojov Kaldi. Vo výsledku bol dosiahnutý WER s hodnotou 16.6\%. Problém klasifikácie textu bol adresovaný s pomocou metód na lineárnu klasifikáciu, konkrétne Multinomial Naive Bayes a Linear Support Vector Machines, kde druhá technika dosiahla vyššiu presnosť klasifikácie.
Comparison of Classification Methods
Dočekal, Martin ; Zendulka, Jaroslav (referee) ; Burgetová, Ivana (advisor)
This thesis deals with a comparison of classification methods. At first, these classification methods based on machine learning are described, then a classifier comparison system is designed and implemented. This thesis also describes some classification tasks and datasets on which the designed system will be tested. The evaluation of classification tasks is done according to standard metrics. In this thesis is presented design and implementation of a classifier that is based on the principle of evolutionary algorithms.
Analysis of Classification Methods
Juríček, Jakub ; Zendulka, Jaroslav (referee) ; Burgetová, Ivana (advisor)
This work deals with the classification methods used in the knowledge discovery from data process and discusses the possibilities of their validation and comparison. Through experiments, the work focuses on the analysis of four selected methods: Naive Bayes classificator, decision tree, neural network and SVM. Factors influencing basic characteristics such as training speed, classification speed, accuracy are examined. A part of the thesis is a desktop application, which is a tool for training, testing and validation of individual methods. Eleven reference data sets are selected for experimental purposes. At the end of this work experimental results of comparison and observed characteristics of classification methods are summarized.
Extension of User Profiles for Targeted Advertising Purposes
Hadač, Filip ; Burgetová, Ivana (referee) ; Zendulka, Jaroslav (advisor)
This thesis is devoted to designation and realisation of the extension of user profiles for improvement targeted advertising purposes. Web scraping is used for acquirement of new data information. Extracted data comes from two servers, ČSFD and Recepty. Data from ČSFD are film genres. Data from Recepty are categories of recepies. Streaming applications are used for processing of data and saving them to databases of user profiles. Preprocessing and machine learning classification algorithms are used for benefit evaluation of new informations for profiles in advertising campaigns. Evaluation of experiments shows that new informations have slight benefit in improvement advertising campaigns.
Automated Detection of Hate Speech and Offensive Language
Štajerová, Alžbeta ; Žmolíková, Kateřina (referee) ; Fajčík, Martin (advisor)
This thesis discusses hate speech and offensive language phenomenon, their respective definitions and their occurrence in natural language. It describes previously used methods of solving the detection. An evaluation of available data sets suitable for the problem of detection is provided. The thesis aims to provide additional methods of solving the detection of this issue and it compares the results of these methods. Five models were selected in total. Two of them are focused on feature extraction and the remaining three are neural network models.  I have experimentally evaluated the success of the implemented models. The results of this thesis allow for comparison of the typical approaches with the methods leveraging the newest findings in terms of machine learning that are used for the classification of hate speech and offensive language.
Monitoring Pedestrian by Drone
Dušek, Vladimír ; Goldmann, Tomáš (referee) ; Drahanský, Martin (advisor)
This thesis is focused on monitoring people in a video footage captured by drone. People are detected by trained model of detector RetinaNet. A feature vector is extracted for each detected person using color histograms. Identification of people is realized by comparing their feature vectors with respect to their distance in the frame. In the end the trajectories of all people are visualized in a panorama image. Accuracy of the trained RetinaNet detector on difficult validation data is 58.6 %. Error rate is partially reduced by the way of algorithm design for trajectory visualisation. It's not necessary to successfully detect person on every frame for correct visualization of its trajectories. At the same time, static objects which are detected as person but are not moving are not consider as people and are not visualized at all. There is a lot of algorithms dealing with people detection however only a few approaches are focused on detection people from an aerial footage.
Music information retrieval techniques for determining the place of origin of the Czech chamber and orchestral music interpretations
Miklánek, Štěpán ; Mekyska, Jiří (referee) ; Kiska, Tomáš (advisor)
This diploma thesis is focused on the statistical analysis of chamber and orchestral classical music recordings composed by Czech authors. One of the chapters is dedicated to the description of a feature extraction process that precedes the statistical analysis. Techniques of Music Information Retrieval are used during several stages of this thesis. Databases used for analysis are described and pre-processing steps are proposed. A tool for synchronization of the recordings was implemented in MATLAB. Finally the system used for classification of recordings based on their geographical origin is proposed. The recordings are sorted by a binary classifier into two categories of Czech and world recordings. The first part of the statistical analysis is focused on individual analysis of features. The features are evaluated based on their discrimination strength. The second part of the statistical analysis is focused on feature selection, which can improve the overall accuracy of the binary classifier compared to the individual analysis of the features.

National Repository of Grey Literature : 412 records found   1 - 10nextend  jump to record:
Interested in being notified about new results for this query?
Subscribe to the RSS feed.