|
|
Machine Comprehension Using Commonsense Knowledge
Daniš, Tomáš ; Landini, Federico Nicolás (referee) ; Fajčík, Martin (advisor)
V tejto práci je skumaná schopnosť používať zdravý rozum v moderných systémoch založených na neurónových sieťach. Zdravým rozumom je myslená schopnosť extrahovať z textu fakty, ktoré nie sú priamo spomenuté, ale implikuje ich situácia v texte. Cieľom práce je poskytnúť náhľad na súčasný stav výskumu v tejto oblasti a nájsť sľubné výskumné smery do budúcnosti. V práci je implementovaný jeden z najmodernejších modelov na odpovedanie na otázky a je ďalej použitý na experimenty v rôznych situáciách. Narozdiel od starších prístupov, tento model dosahuje porovnateľné výsledky s najlepšími známymi modelmi aj keď jeho architektúra neobsahuje žiadne prvky zamerané konkrétne na zlepšenie schopnosti zdravo uvažovať. Taktiež boli nájdené štatistické artefakty v populárnej sade dát s otázkami vyžadujúcimi zdravé uvažovanie. Tieto artefakty môžu byť použité štatistickými modelmi na nájdenie správnej odpovede aj v prípadoch, kedy by to nemalo byť možné. Na základe týchto zistení sú v práci poskytnuté odporúčania a návrhy pre výskum do budúcnosti.
|
|
|
Generating Code from Textual Description of Functionality
Kačur, Ján ; Ondřej, Karel (referee) ; Smrž, Pavel (advisor)
The aim of this thesis was to design and implement system for code generation from textual description of functionality. In total, 2 systems were implemented. One of them served its purpose as a control prototype, the second one was the main product of this thesis. I focused on using smaller non-pre-trained models. Both systems used Transformer type model as their cores. The second system, unlike the first, used syntactic decomposition of both code and textual descriptions. Data used in both systems originated from project CodeSearchNet. Targer programming language to generate was Python. The second system achieved better quantitative results than the first one, with accuracy of 85% versus 60%. The system managed to auto-complete correct code to finish the function definition, with bigger time delay. This thesis is almost exclusively dedicated to the second system.
|
|
|
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.
|
|
|
Named Entity Recognition Exploiting Sub Word Information
Dobrovodský, Patrik ; Egorova, Ekaterina (referee) ; Kesiraju, Santosh (advisor)
Cieľom tejto bakalárskej práce je zhotovenie systému rozpoznania názvoslovnej entity zhotovenej na základe modelu, ktorý bol nedávno považovaný za jeden z najmodernejších a popri tom skúma aký vplyv majú podslovné informácie na nahradenie slov mimo slovnej zásoby. Vytvorený systém vedľa anglického jazyka podporuje aj dva Indo-Európske jazyky konkrétne nemčinu a maďarčinu. Bakalárska práca predstavuje systém využívajúci hlboké učenie pre rozpoznávanie názvoslovných entít, ktorý používa predtrénované a samotrénované slovné vnorenia, zriedkavé vnorenia a charakterové vnorenia vyzdvihnuté konvolučnou neurónovou sieťou. Tieto vnorenia najprv spracujeme sekvenčnou (dlhodobá-krátkodobá pamäť) a potom charakteristickou (podmienené náhodné pole) metódou. Cieľom je dosiahnuť podobnú F1-mieru akú má inšpiračný model s možnosťou porovnania s ostatnými modernými systémami. Výsledkom našej práce je systém, ktorý na anglickej testovacej sade CoNLL 2003 dosiahol 90.98%-né F1-mieru používajúci predtrénované vnorenia a približuje sa k inšpiračnej práci s hodnotou 91.26%. V prípade ďalších jazykov používajúcich samotrénované slovné vnorenia dosiahol systém na testovacej sade WikiAnn pre nemčinu 89.34%-nú a pre maďarčinu 93.04%-nú F1-mieru.
|
|
|
Deep Neural Networks Used for Customer Support Cases Analysis
Marušic, Marek ; Ryšavý, Ondřej (referee) ; Pluskal, Jan (advisor)
Umelá inteligencia je pozoruhodne populárna v dnešnej dobe, pretože si dokáže poradiť s rôznymi veľmi komplexnými úlohami v odvetviach ako napr. spracovanie obrazu, spracovanie zvuku, spracovanie prirodzeného jazyka a podobne. Keďže Red Hat doteraz už vyriešil obrovksé množstvo zákazníckych požiadavkov počas podpory rôznych produktov. Preto bola navrhnutá myšlienka použiť umelú inteligenciu práve na tieto dáta a docieliť tak zlepšenie a zrýchlenie procesu riešenia zákaznícky požiadavkov. V tejto práci sú popísané použité techniky na spracovanie týchto dát a úlohy, ktoré je možné riešiť pomocou hlbokých neurónových sietí. Taktiež sú v tejto práci popísane rôzne modely, ktoré boli vytvorené počas riešenia tejto práce a snažia sa adresovať rôzne úlohy. Ich výkony sú porovnané na spomínaných úlohách.
|
|
|
Word Sense Clustering
Hošták, Viliam Samuel ; Otrusina, Lubomír (referee) ; Smrž, Pavel (advisor)
This thesis deals with semantic similarity of words. It describes and compares existing models that are currently used for this purpose. It discusses the design and implementation of the system for corpus preprocessing, semantic modelling and retrieval of semantically related words. The system that has been created supports the use of distributional semantic models Word2vec, FastText and Glove.
|
|
|
Development of correlation rules for detecting cyber attacks
Dzadíková, Slavomíra ; Safonov, Yehor (referee) ; Martinásek, Zdeněk (advisor)
The diploma thesis deals with the problem of efficient processing of log records and their subsequent analysis using correlation rules. The goal of the thesis was to implement log processing in a structured form, extract individual log fields using a natural language processing model by solving a question answering problem, and develop correlation rules for detecting malicious behavior. Two datasets were produced during the task solution, one with records from Windows devices, and the other containing records from the Fortigate firewall. Pre-trained models based on the BERT and XLNet architecture were created and trained to solve the log parsing problem using the produced datasets, and the results were analyzed and compared. The second part of the thesis was devoted to the development of correlation rules, where the concept of a generic Sigma notation was investigated. It was developed, successfully tested and deployed six correlation rules into own experimental environment in Elastic Stack system. Each rule is also described by tactics, techniques and sub-techniques of the MITRE ATT&CK framework.
|
|
|
Keyword Suggestion in the Central Portal of Czech Libraries
Balaga, Róbert ; Otrusina, Lubomír (referee) ; Smrž, Pavel (advisor)
This thesis deals with various methods of keyphrase extraction from documents, specifically focused on documents from the Central Portal of Czech Libraries. Various methods from statistical, linguistic and graph-based methods have been implemented. Also a new method was suggested, that combines the statistical and linguistic approach. Individual methods have been tested and analyzed according to the standard evaluation metrics, with the suggested method achieving recall of 30 percent.
|
|
|
Automatic Adding of Punctuation into Speech Transcript
Ščavnický, Tomáš ; Veselý, Karel (referee) ; Szőke, Igor (advisor)
This thesis deals with the problem of punctuation reconstruction in the output of automatic speech recognition systems. Constrains given on the solutions were applicability on general spoken English language and reasonable accuracy of the punctuation prediction system. Natural language tends to have in some cases non-deterministic nature and usually consists of a large number of grammatic rules. Therefore, a machine learning approach was chosen to solve this problem for its ability to recognize complicated patterns in data. A number of experiments with recurrent neural networks were executed to find the best network architecture for punctuation prediction. Resulting models created during these experiments reach accuracy comparable if not better than the works currently held as state-of-the-art solutions for punctuation reconstruction.
|
|
|
Comparison of Annotation Tools
Prexta, Dávid ; Otrusina, Lubomír (referee) ; Dytrych, Jaroslav (advisor)
This work deals with the comparison of annotation tools when working with various data sets, and obtaining the results of comparisons useful for improving the knowledge base of the annotators. The thesis analyzes the existing solutions and their drawbacks, from which the proposals of the new solution are deduced. The other sections deals with the design, implementation and testing of the resulting tool, which is evaluated at the conclusion, and possible future extensions are suggested.
|
guest ::
