|
|
Optimization of a Deep Neural Network Label Encoding in a Multi-Label Problem.
Zaťko, Martin ; Novotná, Petra (referee) ; Hejč, Jakub (advisor)
The aim of the diploma thesis is to propose a method of deep learning for the classification of arrhythmias from ECG recordings and to compare the effect of coding its outputs on the overall quality of the model. A 1D convolutional neural network was selected and methods of label coding using one-hot coding, ordinal coding, the method using an autoencoder and the word embbeding method were tested and compared on it. The obtained results show that the use of the word embbeding method can increase the classification capacity of the proposed network.
|
|
|
Detekce kategorie obsahu webové stránky prostřednictvím metod strojového učení.
DOHNAL, Patrik
This bachelor thesis is focused on design and the implementation of the algorithm for classifying the websites into a several categories. The implementation of this software is written in Python. For classifying purposes I use machine learning models such as Naive Bayes classifier, K-Nearest neighbors and Support Vector Machines. Within the process it is assumed to collect my own dataset, wich will be used for training and testing purposes. Thesis also includes detailed description of the methods I uesd.
|
|
|
Optimization of a Deep Neural Network Label Encoding in a Multi-Label Problem.
Zaťko, Martin ; Novotná, Petra (referee) ; Hejč, Jakub (advisor)
The aim of the diploma thesis is to propose a method of deep learning for the classification of arrhythmias from ECG recordings and to compare the effect of coding its outputs on the overall quality of the model. A 1D convolutional neural network was selected and methods of label coding using one-hot coding, ordinal coding, the method using an autoencoder and the word embbeding method were tested and compared on it. The obtained results show that the use of the word embbeding method can increase the classification capacity of the proposed network.
|
|
|
Deep contextualized word embeddings from character language models for neural sequence labeling
Lief, Eric ; Pecina, Pavel (advisor) ; Kocmi, Tom (referee)
A family of Natural Language Processing (NLP) tasks such as part-of- speech (PoS) tagging, Named Entity Recognition (NER), and Multiword Expression (MWE) identification all involve assigning labels to sequences of words in text (sequence labeling). Most modern machine learning approaches to sequence labeling utilize word embeddings, learned representations of text, in which words with similar meanings have similar representations. Quite recently, contextualized word embeddings have garnered much attention because, unlike pretrained context- insensitive embeddings such as word2vec, they are able to capture word meaning in context. In this thesis, I evaluate the performance of different embedding setups (context-sensitive, context-insensitive word, as well as task-specific word, character, lemma, and PoS) on the three abovementioned sequence labeling tasks using a deep learning model (BiLSTM) and Portuguese datasets. v
|
guest ::
