|
|
Trainable Methods for Automatic Biomedical Image Processing
Uher, Václav ; Tučková,, Jana (referee) ; Brezany, Peter (referee) ; Burget, Radim (advisor)
This thesis deals with possibilities of automatic segmentation of biomedical images. For the 3D image segmentation, a deep learning method has been proposed. In the work problems of network design, memory optimization method and subsequent composition of the resulting image are solved. The uniqueness of the method lies in 3D image processing on a GPU in combination with augmentation of training data and preservation of the output size with the original image. This is achieved by dividing the image into smaller parts with the overlay and then folding to the original size. The functionality of the method is verified on the segmentation of human brain tissue on magnetic resonance imaging, where it overcomes human accuracy when compared a specialist vs. specialist, and cell segmentation on a slices of the Drosophila brain from an electron microscope, where published results from the impacted paper are overcome.
|
|
|
Klasifikace na množinách bodů v 3D
Střelský, Jakub ; Mráz, František (advisor) ; Šikudová, Elena (referee)
Increasing interest for classification of 3D geometrical data has led to discov- ery of PointNet, which is a neural network architecture capable of processing un- ordered point sets. This thesis explores several methods of utilizing conventional point features within PointNet and their impact on classification. Classification performance of the presented methods was experimentally evaluated and com- pared with a baseline PointNet model on four different datasets. The results of the experiments suggest that some of the considered features can improve clas- sification effectiveness of PointNet on difficult datasets with objects that are not aligned into canonical orientation. In particular, the well known spin image rep- resentations can be employed successfully and reliably within PointNet. Further- more, a feature-based alternative to spatial transformer, which is a sub-network of PointNet responsible for aligning misaligned objects into canonical orientation, have been introduced. Additional experiments demonstrate that the alternative might be competitive with spatial transformer on challenging datasets. 1
|
|
|
Adversarial examples generation for deep neural networks
Procházka, Štěpán ; Neruda, Roman (advisor) ; Kratochvíl, Miroslav (referee)
Machine learning models exhibit vulnerability to adversarial examples i.e., artificially created inputs that become misinterpreted. The goal of this work is to explore black-box adversarial attacks on deep networks performing image classification. The role of surrogate machine learning models for adversarial attacks is studied, and a special version of a genetic algorithm for generating adversarial examples is proposed. The efficiency of attacks is validated by a multitude of experiments with the Fashion MNIST data set. The experimental results verify the usability of our approach with surprisingly good performance in several cases, such as non-targeted attack on residual networks.
|
|
|
Using DNN for triangular network analysis in geometric morphometry
Dvořáková, Gabriela ; Pelikán, Josef (advisor) ; Dupej, Ján (referee)
The aim of this thesis is to use deep learning for the task of 3D object recognition. Deep learning has been succesfully used for three dimensional data recognition. However, most of the published work chose to represent 3D objects as a set of projected 2D pixel images or in the form of binary voxels. The main goal is to propose an alternative mapping of 3D data to the NN input. Three data representations are introduced: Treating vertex coordinates as a 1D array, projection to a 2D grid and a set of surface oblique lines crossing the sig- nificant parts of an object. All of the proposed data representations are tested for the gender classification task using NN and CNN on 3D facial models. We analyzed the impact of coordinate relativization and a new modified dataset crea- ted by extracting a nose area from original triangle meshes. Experimental results confirmed the quality of the oblique lines approach with achieved classification accuracies of 84, 2% using CNN. 1
|
|
|
Navigation Using Deep Convolutional Networks
Skácel, Dalibor ; Veľas, Martin (referee) ; Hradiš, Michal (advisor)
In this thesis I deal with the problem of navigation and autonomous driving using convolutional neural networks. I focus on the main approaches utilizing sensory inputs described in literature and the theory of neural networks, imitation and reinforcement learning. I also discuss the tools and methods applicable to driving systems. I created two deep learning models for autonomous driving in simulated environment. These models use the Dataset Aggregation and Deep Deterministic Policy Gradient algorithms. I tested the created models in the TORCS car racing simulator and compared the result with available sources.
|
|
|
A Library for Convolutional Neural Network Design
Rek, Petr ; Mrázek, Vojtěch (referee) ; Sekanina, Lukáš (advisor)
In this diploma thesis, the reader is introduced to artificial neural networks and convolutional neural networks. Based on that, the design and implementation of a new library for convolutional neural networks is described. The library is then evaluated on widely used datasets and compared to other publicly available libraries. The added benefit of the library, that makes it unique, is its independence on data types. Each layer may contain up to three independent data types - for weights, for inference and for training. For the purpose of evaluating this feature, a data type with fixed point representation is also part of the library. The effects of this representation on trained net accuracy are put to a test.
|
|
|
Improving Bots Playing Starcraft II Game in PySC2 Environment
Krušina, Jan ; Škoda, Petr (referee) ; Smrž, Pavel (advisor)
The aim of this thesis is to create an automated system for playing a real-time strategy game Starcraft II. Learning from replays via supervised learning and reinforcement learning techniques are used for improving bot's behavior. The proposed system should be capable of playing the whole game utilizing PySC2 framework for machine learning. Performance of the bot is evaluated against the built-in scripted AI in the game.
|
|
|
A convolutional neural network for image segmentation
Mitrenga, Michal ; Petyovský, Petr (referee) ; Jirsík, Václav (advisor)
The aim of the bachelor thesis is to learn more about the problem of convolutional neural networks and to realize image segmentation. This theme includes the field of computer vision, which is used in systems of artificial intelligence. Special Attention is paid to the image segmentation process. Furthermore, the thesis deals with the basic principles of artificial neural networks, the structure of convolutional neural networks and especially with the description of individual semantic segmentation architectures. The chosen SegNet architecture is used in a practical application along with a pre-learned network. Part of the work is a database of CamVid images, which is used for training. For testing, a custom image database is created. Practical part is focused on CNN training and searching for unsuitable parameters for network learning using SW Matlab.
|
|
|
Artificial Composition of Multi-Instrumental Polyphonic Music
Samuel, David ; Pilát, Martin (advisor) ; Neruda, Roman (referee)
David Samuel We propose a generative model for artificial composition of both classical and popular music with the goal of producing music as well as humans do. The problem is that music is based on a highly sophisticated hierarchical structure and it is hard to measure its quality automatically. Contrary to other's work, we try to generate a symbolic representation of music with multiple different instruments playing simultaneously to cover a broader musical space. We train three modules based on LSTM networks to generate the music; a lot of effort is put into reducing high complexity of multi-instrumental music representation by a thorough musical analysis. Our work serves mainly as a proof-of-concept for music composition. We believe that the proposed preprocessing techniques and symbolic representation constitute a useful resource for future research in this field. 1
|
|
|
Convolutional neural networks for identification of axial 2D slices in CT data
Vavřinová, Pavlína ; Harabiš, Vratislav (referee) ; Jakubíček, Roman (advisor)
This thesis deals with the classification of axial 2D slices in CT patient’s data into six categories. The sphere of convolutional neural networks was used for this purpose. For a better understanding of this issue, the basics of neural networks and then the principles of deep learning including convolutional neural networks are explained at first. The AlexNet network was specifically selected for the intention of this identification, and it was tested on the created data set after being adaptated. The overall classification success rate was 86% ,after the final adjustments, a slight improvement was achieved and the identification success rate was 87%.
|
guest ::
