|
|
Rozpoznání hranic jízdního pruhu v záběrech palubní kamery
Letovanec, Lukáš ; Bartl, Vojtěch (oponent) ; Herout, Adam (vedoucí práce)
Táto práca sa zaoberá problematikou rozpoznávania hraníc jazdného pruhu v záberoch palubnej kamery. V práci je predstavená architektúra hlbokej konvolučnej neurónovej siete, pomocou ktorej sa rieši spomínaný problém. Sieť bola trénovaná na rozsiahlej dátovej sade pomocou algoritmu gradientového zostupu. Natrénovaný model preukázal schopnosť kvalitne rozpoznávať hranice jazdného pruhu v rôznych situáciách a podmienkach. Výsledok práce potvrdzuje, že hlboké konvolučné neurónové siete sú vhodným nástrojom pre rozpoznávanie hraníc jazdného pruhu.
|
|
|
Traffic analysis using on machine learning
Zelený, Ondřej ; Slanina, Martin (oponent) ; Frýza, Tomáš (vedoucí práce)
The aim of this semestral thesis is to investigate the problematic of object detection and clasification for traffic analysis. The theoretical part of the paper takes insight on numerous methods and techniques of object detection and clasification. Further the paper discuses popular frameworks and programming languages for implementation of convolutional neural networks. The practical part shows implementation of chosen model and hardware selection for real system.
|
|
|
Pilot proficiency classification from gaze
Ruta, Dominik ; Vlk, Jan (oponent) ; Chudý, Peter (vedoucí práce)
This work deals with the classification of pilot proficiency level and basic flight maneuvers from gaze. The goal is to provide additional valuable tool for aviation instructors to evaluate proficiency of pilot students and provides them with feedback. This idea is based on results of numerous relevant studies, which discovered correlation between effective scanning patterns and domain performance. This thesis considers two proficiency levels~---~amateur and experienced. This work utilizes common analysis metrics of visual scanning and machine-learning classification techniques. The Support Vector Machine algorithm is used for the proficiency classification and Hidden Markov Models are utilized in basic flight maneuvers classification. The result of this thesis is a high accuracy proficiency classification and good ability to distinguish between individual basic flight maneuvers performed by pilots.
|
|
| |
|
|
Active Learning pro zpracování archivních pramenů
Hříbek, David ; Zbořil, František (oponent) ; Rozman, Jaroslav (vedoucí práce)
V teto praci je reseno vytvoreni systemu, ktery umoznuje nahrani a anotaci skenu historickych dokumentu a nasledne aktivni doucovani modelu pro rozpoznavani znaku (OCR) na dostupnych anotacich (vyznacenych radcich a jejich prepisech). V praci je popsan proces, klasifikovany techniky a uveden existujici system pro rozpoznavani znaku. Predevsim je kladen duraz na metody strojoveho uceni. Dale jsou vysvetleny metody aktivniho uceni a navrhnut zpusob doucovani OCR modelu z anotovanych skenu. Zbytek prace se zabyva konkretnim navrhem, implementaci, dostupnymi datasety, vyhodnocenim uspesnosti rozpoznavani znaku vlastnorucne vytvoreneho OCR modelu a testovanim celeho systemu.
|
|
|
Identification and characterization of malicious behavior in behavioral graphs
Varga, Adam ; Burget, Radim (oponent) ; Hajný, Jan (vedoucí práce)
In recent years, there has been an increase in work involving comprehensive malware detection. It is often useful to use a graph format to capture behavior. This is the case with the Avast antivirus program, whose behavioral shield detects malicious behavior and stores it in the form of graphs. Since this is a proprietary solution and Avast antivirus works with its own set of characterized behavior, it was necessary to design our own detection method that will be built on top of these behavioral graphs. This work analyzes graphs of malware behavior captured by the behavioral shield of the Avast antivirus program for the process of deeper detection of malware. Detection of malicious behavior begins with the analysis and abstraction of patterns from the behavioral graph. Isolated patterns can more effectively identify dynamically changing malware. Behavior graphs are stored in the Neo4j graph database and thousands of them are captured every day. The aim of this work was to design an algorithm to identify the behavior of malicious software with emphasis on tagging speed and uniqueness of identified patterns of behavior. Identification of malicious behavior consists in finding the most important properties of trained classifiers and subsequent extraction of a subgraph consisting only of these important properties of nodes and the relationships between them. Subsequently, a rule for the evaluation of the extracted subgraph is proposed. The diploma thesis took place in cooperation with Avast Software s.r.o.
|
|
|
Strojové učení v přirozeném jazyce
Otrusina, Lubomír ; Šilhavá, Jana (oponent) ; Smrž, Pavel (vedoucí práce)
Tato práce se zabývá zjednoznačňováním slovních významů pomocí metod strojového učení. Čtenář je krátce seznámen s danou problematikou a jejím historickým vývojem. Jsou zde popsány nejpoužívanější metody a přístupy, speciálně pak naivní Bayesův klasifikátor, který je implementován v systému. Je zde uveden i názorný příklad pro tento klasifikátor. V praktické části je popsán návrh systému využívající tohoto klasifikátoru včetně popisu různých algoritmů použitých v systému. Na závěr je uvedeno vyhodnocení výsledků systému a jejich analýza. Implementovaný systém se zúčastnil soutěže v rámci mezinárodní konfernce sémantického vyhodnocování SemEval-2007.
|
|
|
Optimization of DDoS Mitigation Rule Inference
Carasec, Elena ; Grégr, Matěj (oponent) ; Žádník, Martin (vedoucí práce)
This thesis discusses the possibility of using machine learning algorithms for DDoS protection. For classical and incremental (online) learning are considered explainable supervised learning methods, particularly decision trees. Furthermore, some possible optimisations are introduced to increase traffic classification accuracy and decrease the amount of blocked legitimate traffic.
|
|
|
Traffic analysis using on machine learning
Zelený, Ondřej ; Slanina, Martin (oponent) ; Frýza, Tomáš (vedoucí práce)
The aim of this thesis is to investigate the problematic of object detection and classification for traffic analysis. The theoretical part of the paper takes insight on numerous methods and techniques of object detection and classification. Further the paper discuses popular frameworks and programming languages for implementation of convolutional neural networks as well as multi-object tracking and communication with IoT server. The practical part shows implementation of chosen model and additional functionalities, object trackers and communication with selected IoT platform as well as data processing in cloud and visualization.
|
|
|
Traffic analysis using on machine learning
Zelený, Ondřej ; Slanina, Martin (oponent) ; Frýza, Tomáš (vedoucí práce)
The aim of this thesis is to investigate the problematic of object detection and classification for traffic analysis. The theoretical part of the paper takes insight on numerous methods and techniques of object detection and classification. Further the paper discuses popular frameworks and programming languages for implementation of convolutional neural networks as well as multi-object tracking and communication with IoT server. The practical part shows implementation of chosen model and additional functionalities, object trackers and communication with selected IoT platform as well as data processing in cloud and visualization.
|
host ::
RSS.