|
|
Eluding and Evasion of IDS Systems
Černý, Marek ; Tobola, Jiří (referee) ; Žádník, Martin (advisor)
This paper analyzes network security devices called intrusion detection (ID) systems. In order to point out possible flaws, especially ID systems using signature analysis are examined. Based on this, methods to exploit possible vulnerabilities of these systems were designed. These methods were implemented into a simple program for ID systems efficiency evaluation. It can be used in a way entirely independent of particular network attack used in the test.
|
|
|
IPv6 Modelling in OMNeT++
Černý, Marek ; Grégr, Matěj (referee) ; Veselý, Vladimír (advisor)
OMNeT++ is a discrete-event simulator commonly used to build various network simulations. It can be extended by INET framework that supplies models of protocols and devices from TCP/IP environment. This paper explores current capabilities of INET to model IPv6, particularly routing. Implemented extension includes modules of dual-stack router and dual-stack host that support XML-based configuration of IPv6 addressing and static routing. We also introduce an OSPFv3 module that implements most of auxiliary features and is ready for future development of the routing protocol core.
|
|
|
Production of "flange" components on CNC machines
Černý, Marek ; Trčka, Tomáš (referee) ; Kalivoda, Milan (advisor)
This bachelor thesis deals with the production of flange type components on CNC machines. The flange function in the tank assembly is described here. The work includes choice of material, choice of semifinished product and proposal of machinery for production. Also, there is a technological process and NC programs for various control systems. Finally, a prototype is produced, which is then evaluated.
|
|
|
Machining processes in furniture industry
Černý, Marek ; Kalivoda, Milan (referee) ; Chladil, Josef (advisor)
This master thesis deals with machining processes in the furniture industry. Various group of materials, which are used in furniture industry are described there. Then there are described assembly and machining processes, such as manual and machine. In practical part, the production process for the selected sample product was proposed. The production was performed in a specific company, which has suitable machinery and tools. Sample product was made from double-sided laminated medium density fibreboard, which belongs to the group of wood-based materials. As a part of the debugging produciton process, the dependence tool life at cutting speed was experimentaly found. The experiment was performed on CNC milling machine with cutting speeds in the range 453 to 905 m.min-1, width draft of cutting edge 9,5 mm and feed per tooth 0,05 mm. Analog passametr Somet with an accuracy 0,001 mm was used to measure tool wear. The measured data were recorded and evaluated in program Excel. The resulting dependency tool life at cutting speed can be used in other machining operations. After performing the experiment, each parts of the product were produced and mountig was performed. The conclusion of the thesis contains an evaluation of the proposed production process.
|
|
|
Synthesis of higher order passive RLC filters for educational laboratory purposes
Černý, Marek ; Petržela, Jiří (referee) ; Rujzl, Miroslav (advisor)
This bachelor thesis deals with the Synthesis of higher order passive RLC filters. The main target was to design a set of filters for educational laboratory purposes, which will contain filters of different orders and approximations (Chebyshev, Inverse Chebyshev and Cauer). The filters were designed according to design tables for standardized lowpass filters and were subsequently simulated in the PSpice program (Frequency analysis, Monte Carlo, Worst Case). Subsequently, a laboratory preparation was designed and revived. Individual filters were tested and compared with ideal waveforms.
|
|
|
Differentiable Depth Estimation for Bin Picking
Černý, Marek ; Klusáček, David (advisor) ; Šikudová, Elena (referee)
The goal of this thesis was to investigate the neural 3D surface reconstruction from multiple views with the intent to use the resulting depth maps for bin picking. Survey of papers from 2014 to 2018 showed that none of the state of the art methods would be used to control a robot arm in our setup. Therefore we decided to create our low-level neural approach which we called the EmfNet. The network is based on a pyramidal resolution refining approach. At each pyramid's layer, there are three separate networks that take part in the computation. Each of them has a definite goal, which gives us almost complete understanding of what is going on inside the network. The EmfNet model was partially usable, but we nevertheless extended it to EmfNet-v2. First, another measuring layer was added, which freed EmfNet from depending on an unnecessary hyperparameter. Second, we used constraints on geometry for the network not to be confused by occlusions (cases where a certain part of the surface is visible only from a single camera). Both networks were implemented and tested on a corpus that was created as a part of this thesis. A corpus containing rendered as well as real data. The process of correspondence pairing inside the network can be observed using the visualization tool. We designed a way how to use a robotic arm...
|
|
|
Machining processes in furniture industry
Černý, Marek ; Kalivoda, Milan (referee) ; Chladil, Josef (advisor)
This master thesis deals with machining processes in the furniture industry. Various group of materials, which are used in furniture industry are described there. Then there are described assembly and machining processes, such as manual and machine. In practical part, the production process for the selected sample product was proposed. The production was performed in a specific company, which has suitable machinery and tools. Sample product was made from double-sided laminated medium density fibreboard, which belongs to the group of wood-based materials. As a part of the debugging produciton process, the dependence tool life at cutting speed was experimentaly found. The experiment was performed on CNC milling machine with cutting speeds in the range 453 to 905 m.min-1, width draft of cutting edge 9,5 mm and feed per tooth 0,05 mm. Analog passametr Somet with an accuracy 0,001 mm was used to measure tool wear. The measured data were recorded and evaluated in program Excel. The resulting dependency tool life at cutting speed can be used in other machining operations. After performing the experiment, each parts of the product were produced and mountig was performed. The conclusion of the thesis contains an evaluation of the proposed production process.
|
|
|
Persistent homology and neural networks
Černý, Marek ; Šámal, Robert (advisor) ; Tancer, Martin (referee)
Deep learning has solved various computer vision problems in the last decade. We use computational topology, namely persistent homology groups, to analyze the spaces of internal feature representations of convolutional neural networks (CNNs). We observed the correspondence between the summaries of the homological persistence groups of the decision boundary and ResNet-18 CNN training error during the phenomenon of deep double descent. Furthermore, a considered specific persistence summary suggests an analogy to epoch-wise double descent so that we may better understand the internal representations during the critically parametrized regime. Mainly, we develop and compare multiple approaches to CNN models processing using the persistence homology. Our best method, the Differentiable Persistence Accuracy Es- timator (DPAE), achieves very high accuracy in predicting CNN's performance (R2 score close to 0.99). Our DPAE is a topologically optimized end-to-end architecture involving differentiable persistence computation. Moreover, DPAE overperforms the original clas- sical machine learning-based method on its native Small CNN Zoo dataset. We publicly release the complete source code of DPAE and other presented experiments. 1
|
|
|
Differentiable Depth Estimation for Bin Picking
Černý, Marek ; Klusáček, David (advisor) ; Šikudová, Elena (referee)
The goal of this thesis was to investigate the neural 3D surface reconstruction from multiple views with the intent to use the resulting depth maps for bin picking. Survey of papers from 2014 to 2018 showed that none of the state of the art methods would be used to control a robot arm in our setup. Therefore we decided to create our low-level neural approach which we called the EmfNet. The network is based on a pyramidal resolution refining approach. At each pyramid's layer, there are three separate networks that take part in the computation. Each of them has a definite goal, which gives us almost complete understanding of what is going on inside the network. The EmfNet model was partially usable, but we nevertheless extended it to EmfNet-v2. First, another measuring layer was added, which freed EmfNet from depending on an unnecessary hyperparameter. Second, we used constraints on geometry for the network not to be confused by occlusions (cases where a certain part of the surface is visible only from a single camera). Both networks were implemented and tested on a corpus that was created as a part of this thesis. A corpus containing rendered as well as real data. The process of correspondence pairing inside the network can be observed using the visualization tool. We designed a way how to use a robotic arm...
|
|
|
Production of "flange" components on CNC machines
Černý, Marek ; Trčka, Tomáš (referee) ; Kalivoda, Milan (advisor)
This bachelor thesis deals with the production of flange type components on CNC machines. The flange function in the tank assembly is described here. The work includes choice of material, choice of semifinished product and proposal of machinery for production. Also, there is a technological process and NC programs for various control systems. Finally, a prototype is produced, which is then evaluated.
|
guest ::
RSS feed.