Přístupnostní navigace
E-application
Search Search Close
study programme
Faculty: FITAbbreviation: DIT-ENAcad. year: 2024/2025
Type of study programme: Doctoral
Study programme code: P0613D140029
Degree awarded: Ph.D.
Language of instruction: English
Accreditation: 8.12.2020 - 8.12.2030
Profile of the programme
Academically oriented
Mode of study
Full-time study
Standard study length
4 years
Programme supervisor
prof. Ing. Lukáš Sekanina, Ph.D.
Doctoral Board
Chairman :prof. Ing. Lukáš Sekanina, Ph.D.Councillor internal :doc. Ing. Jan Kořenek, Ph.D.prof. Dr. Ing. Jan Černockýprof. Ing. Adam Herout, Ph.D.prof. Ing. Tomáš Hruška, CSc.doc. Dr. Ing. Petr Hanáčekprof. Dr. Ing. Pavel Zemčík, dr. h. c.prof. RNDr. Alexandr Meduna, CSc.prof. Dr. Ing. Zbyněk Raidaprof. RNDr. Josef Šlapal, CSc.prof. Ing. Pavel Václavek, Ph.D.prof. Ing. Tomáš Vojnar, Ph.D.Councillor external :prof.,RNDr. Jiří Barnat, Ph.D.
Fields of education
Study aims
The goal of the doctoral degree programme is to provide outstanding graduates from the master degree programme with a specialised university education of the highest level in certain fields of computer science and information technology, including especially the areas of information systems, computer-based systems and computer networks, computer graphics and multimedia, and intelligent systems. The education obtained within this degree programme also comprises a training and attestation for scientific work.
Graduate profile
Profession characteristics
FIT graduates in general and FIT doctoral graduates in particular do not have a problem finding employment at scientific, pedagogical or management positions both in Czech Republic and abroad.
Fulfilment criteria
The requirements that the doctoral students have to fulfil are given by their individual study plans, which specify the courses that they have to complete, their presupposed study visits and active participation at scientific conferences, and their minimum pedagogical activities within the bachelor and master degree programmes of the faculty. A successful completion of the doctoral studies is conditional on the following:
Study plan creation
The rules are determined by the directions of the dean for preparing the individual study plan of a doctoral student. The plan is to be based on the theme of his/her future dissertation thesis and it is to be approved by the board of the branch.
https://www.fit.vut.cz/fit/info/smernice/sm2018-13-en.pdf
Availability for the disabled
Brno university of technology provides studies for persons with health disabilities according to section 21 par. 1 e) of the Act no. 111/1998, about universities and about the change and supplementing other laws (Higher Education Act) as amended, and according to the requirements in this field arising from Government Regulation No. 274/2016 Coll., on standards for accreditation in higher education, provides services for study applicants and students with specific needs within the scope and in form corresponding with the specification stated in Annex III to Rules for allocation of a financial contribution and funding for public universities by the Ministry of Education, Youth and Sports, specifying financing additional costs of studies for students with specific needs.Services for students with specific needs at BUT are carried out through the activities of specialized workplace - Alfons counselling center, which is a part of BUT Lifelong Learning Institute - Student counselling section.Counselling center activities and rules for making studies accessible are guaranteed by the university through a valid Rector's directive 11/2017 concerning the status of study applicants and students with specific needs at BUT. This internal standard guarantees minimal stadards of provided services.Services of the counselling center are offered to all study applicants and students with any and all types of health disabilities stated in the Methodological standard of the Ministry of Education, Youth and Sports.
What degree programme types may have preceded
The study programme builds on both the ongoing follow-up Master's programme in Information Technology and the new follow-up Master's programme in Information Technology and Artificial Intelligence.Students can also, according to their needs and outside their formalized studies, take courses and trainings related to the methodology of scientific work, publishing and citation skills, ethics, pedagogy and soft skills organized by BUT or other institutions.
Issued topics of Doctoral Study Program
Topic Description:
The current Internet trend is to move applications, data, and computing to the cloud. This shift is affecting individuals as well as enterprise environments. This trend challenges the way applications and services are monitored, as traditional monitoring techniques such as Netflow or SNMP are unable to monitor the performance, availability and responsiveness of cloud applications. In addition, the availability of monitoring and diagnostic information from the cloud is limited and often dependent on the type of cloud and service availability.
This research topic focuses on designing active monitoring methods for cloud applications using a network of monitoring agents that monitor the availability, performance, and security of applications in the cloud at the L3-L7 layers using modular tests. The monitoring data will be processed using machine learning methods with a focus on behavioral profiling, predictive analysis and fault detection.
References:
Tutor: Matoušek Petr, doc. Ing., Ph.D., M.A.
The topic concerns algorithms of image, video, and/or signal processing. Its main goal is to research and in-depth analyze existing algorithms and discover new ones so that they have desirable features and so that they are possible to efficiently implement. Such efficient implementation can be but does not necessarily have to be part of the work but it is important to prepare the algorithms so that they can be efficiently implemented e.g. in CPU, in CPU with acceleration through SSE instructions, in embeded systems, even in combination with FPGA, in Intel Xeon PHI, in extremely low power systems, or in other environments. It is possible to exploit algorithms of artificial intelligence, such as neural networks, especially CNNs The application possibilities of the algorithms are also important and the application can be but does not have to be part of the work. The algorithms/applications of interest include:
After mutual agreement, individually selected algorithms can be considered as well as soon as they do belong to the general topic.
Collaboration on grant projects, such as TACR, MPO, H2020, ECSEL (possible employment or scholarship).
Tutor: Zemčík Pavel, prof. Dr. Ing., dr. h. c.
The project is concerned with advanced methods of computational photography. The aim is to research new computational photography methods, which comprises software solutions potentially supported by new optics and/or hardware. Our interest is on HDR image and video processing, color-to-grayscale conversions, spectral imaging, generative AI and others.
Tutor: Čadík Martin, doc. Ing., Ph.D.
Předmětem studia budou algoritmy zobrazování scén v reálném čase s využitím moderních grafických procesorů. Práce se má zaměřit na složité a rozsáhlé scény s různými materiály a množstvím světelných zdrojů. Výsledkem práce by měly být algoritmy umožňující maximálně využít moderní grafické procesory - a to algoritmy použitelné v programovatelných částech zobrazovacího řetězce i algoritmy využívající GPU jako obecný vysoce paralelní stroj (architektury CUDA a OpenCL).
Tutor: Herout Adam, prof. Ing., Ph.D.
The project is concerned with advanced rendering and global illumination methods. The aim is to research new photorealistic (physically accurate) as well as non-photorealistic (NPR) simulations of interaction of light with the 3D scene. Cooperation and research visits with leading research labs are possible (Adobe, USA, MPII Saarbrücken, Německo, Disney Curych, Švýcarsko, INRIA Bordeaux, Francie).
The usage of large pre-trained models has become ubiquitous in several fields of Artificial Intelligence (AI). The recent developments and capabilities of large language models are a prime example. Similar trends are seen in areas such as speech technology, computer vision, and across disciplines related to medicine and healthcare. In speech and language processing, current state-of-the-art models are trained independent of each other, and a majority of them are uni-modal at their input. Whereas, a number of applications such as spoken language translation, task-oriented dialogue systems and atypical speech assessment either require or benefit from a careful combination of two or more models. A naive way of building a cascade pipeline results in error propagation and compounding, while joint-training causes catastrophic forgetting, where the benefits of pre-training diminish. Combined with these limitations, the black-box nature of the models make them hard to interpret; moreover, they propagate harmful biases acquired from the massive web-crawled training data. To overcome these limitations of the current state-of-the-art, this PhD topic aims to develop theoretically-motivated methods for aligning any arbitrary pre-trained models via an interpretable latent space. The alignment will enable to join the models without requiring to fine-tune them. The interpretable latent space will ease the study and identification of the linguistic, para-linguistic, and fairness attributes that are encoded in the pre-trained models. This will also allow the explainability of the models' output in human-centred applications related to medicine and healthcare such as atypical speech and language assessment. The shared latent space enables to use efficient data augmentation and bias mitigation methods that will enhance the robustness of speech and language applications.
Tutor: Černocký Jan, prof. Dr. Ing.
The dissertation focuses on the security of wireless local area networks. The student should become familiar with selected wireless networks and their security as part of the solution. This work aims to study the theory of wireless networks, their properties, and possibilities of attacks, test the basic types of attacks, design new protection methods, conduct experiments, evaluate the results, and propose the direction of further research.
Participation in relevant international conferences and publication in scientific journals are expected.
Co-supervised by Dr. Kamil Malinka.
Tutor: Hanáček Petr, doc. Dr. Ing.
Transformers are large language models based on deep learning. They are mainly used for natural language processing, but also for document analysis, information retrieval, etc.
The research topic focuses on the application of large LLM-type language models for information retrieval in technical documentation, e.g. technical reports, manuals, domain-specific knowledge bases, etc. The research involves processing the input domain-oriented documents and converting them into a language model to be used in transfer learning technique.
The goal of the research is to apply and optimize transformers for efficient retrieval of domain-specific information, e.g. to support network administrators in handling security incidents, network diagnostics, etc.
Problem Statement: Mental stress, anxiety and depression are mental health conditions that often occur together. In such a case, the person is stressed and is not able to control the worry, and it correspondingly affects his/her social and occupational activities. Hence, proper assessment and diagnosis for mental stress, anxiety and depression is required in order for a person to effectively keep taking part in his/her normal daily tasks and activities.
Issues with Current Solutions: Unfortunately, conventional assessment and diagnostic measures are subjective in nature and are used only when the symptoms are already evident due to advanced stages of mental stress, anxiety and depression. However, mental stress, anxiety and depression do not occur overnight, rather it is a long process. Hence, detection of symptoms is required at early stages of mental stress, anxiety and depression because that may result in a cure or at least it will delay the onset of serious mental health issues associated with them.
Challenges: Unlike other diseases where the symptoms like fever and cough allow people to seek help, symptoms at early stages of mental stress and anxiety are not easily identifiable. Hence, the brain needs to be continuously monitored for any sign of change or deterioration in order to detect the symptoms at early stage.
Solution: The solution lies in the development of an objective and quantitative method that can detect mental stress, anxiety and depression at an early stage. Perception of mental stress, anxiety and depression originates in the brain; therefore, this research investigates the neurophysiological features extracted from brain electroencephalogram (EEG) signal to measure mental stress, anxiety and depression at early stage. This will require development of method for extraction of features as well as pattern recognition approach to provide a solution. The EEG dataset is already available for this project.
Few Words About Supervision: I have extensive experience of working in the field of neuro-signal and neuroimage processing and I am currently head of a research group in this area. This is a multidisciplinary project and it will involve working with clinicians. However, the core of the project is related to IT in terms of development of a new method. Please feel free to contact me at malik@fit.vutbr.cz
Tutor: Malik Aamir Saeed, prof., Ph.D.
Language is the basic linking element of any nation and its territorial dialects are an important part of regional identity. The PhD topic, proposed by specialists in ASR (BUT), dialectology (ÚJČ) and interactive mapping (UPOL), focuses on the adaptation of existing technologies and development of new procedures for automatic processing, preservation, documentation and presentation of the dialects of the Czech language. The primary goal is to create a powerful ASR system usable by dialectologists, based on large pre-trained models, datasets of a specific language and/or related languages, and small amounts of data from target dialect areas. The topic is related to the work on the JAMAP project. . .
Tutor: Burget Lukáš, doc. Ing., Ph.D.
Cílem disertační práce je výzkum v oblasti tzv. generativní umělé inteligence - ať už se jedná o difusní a adversiální modely pro generování videa, generativní textové modely pro vytváření příběhů, automatické generování počítačového kódu, hudby, reprezentaci znalostí fyziky a chemie a podporu vědecké kreativity, případně kombinace všech těchto přístupů. Práce se zaměří na řešení problémů interakce člověka s generovanými mezivýsledky, přirozeného označování jednotlivých částí a konceptů tak, aby bylo možné na průběžné výsledky navazovat, a na vývoj metod úpravy datových sad a postupů učení, aby bylo možné řešit společenské problémy, spojené s vytvářenými kreativními modely - otázky spravedlivosti modelů, předpojatosti a začlenění konceptů tzv. zodpovědné umělé inteligence.
Tutor: Smrž Pavel, doc. RNDr., Ph.D.
Konverzační agenti se pomalu stávají běžnou součástí rozhraní pro (prvotní) komunikaci se zákazníkem a odpovídání na jeho otázky. Výzkum v oblasti počítačového zpracování přirozeného jazyka se zaměřuje na vytvoření automatické klasifikace prvotní komunikace a, zejména, otázky uživatele, do předem daných tříd, k nimž existují konkrétní texty. Není však uspokojivě vyřešeno rozšiřování "znalostí" komunikačních agentů při aktualizaci strukturovaných dat, případně při přidání dalších textových materiálů.Cílem disertační práce je rozvinout existující přístupy využívající obrovské kolekce neanotovaných textových dat a způsoby kombinování strukturované a nestrukturované znalosti a optimalizace procesů při rozšiřování funkcionality stávajících i nových konverzačních agentů. Součástí práce bude i aplikace zkoumaných metod v rámci evropských projektů, na jejichž řešení se školitel podílí.
The dissertation focuses on the security of IoT systems. This work aims to study the theory of IoT systems, their properties, and possibilities of attacks, test the basic types of attacks, design a new method of protection, conduct experiments, evaluate results, and design further research.
Critical infrastructure consists of systems and elements whose failure would have a serious impact on the provision of basic services to the public, such as water, electricity or gas distribution. Examples of critical infrastructure assets include power plants, substations, water facilities, gas distribution, traffic control, etc. These systems use industrial ICS control protocols such as Modbus, IEC 104, MMS for communication.
The research includes the analysis of cyber threats in industrial communications according to MITRE ATT&CK for ICS and the design of an anomaly detection system. Based on the analysis of normal traffic, attributes representing ICS communication must be automatically selected to build an anomaly detection model, which can be built using formal languages, statistical methods, machine learning methods, or neural networks. These models describe the expected behavior of the system and are used to detect anomalies. When an anomaly is detected, the proposed system evaluates its severity, determines its cause and method of resolution using the knowledge base.
The work will start with getting familiar with the basics of the problem of voice deep fake detection (DFD), terminology, available techniques, data and challenges (especially AVSpoof), with the history and state of the art techniques and tools for speaker recognition (wespeaker toolkit), with state of the art techniques and tools for personalized text to speech (pTTS) synthesis and voice conversion. The first task will be reproducing one or two DF detection systems from AVSpoof 2021 (or a newer challenge), checking that the numbers match what is reported, studying how the systems work, followed by attacking the AVSpoof 2021 DFD system(s) with several up to date DF creation techniques. The main task of the PhD work is to suggest and implement ways to detect DFD (or help DFD detection) by for example (1) making the DFD system aware of genuine speech of the target speaker (2) work on artifacts that might be badly handled by pTTS systems, such as breaths. (3) suggesting and implementing techniques making use of psychoacoustical findings (4) suggesting and implementing techniques making use of text information available from the target speaker (such as social media).
Problem Statement: Alcohol addiction is a chronic and complex brain disorder causing devastating individual and social problems. Additionally, alcohol causes 3.3 million deaths a year worldwide, close to 6% of all deaths. Many of these deaths are associated with alcohol addiction. Therefore, it's important to look into methods for the diagnosis as well as the treatment of alcohol addiction.
Issues with Current Solutions: Conventionally, screening and assessment of alcohol-related problems are mainly based on self-test reports. However, the accuracy of self-test reports has been questioned, especially for heavy drinkers, because the self-test reports may misguide the diagnosis due to the patient's memory loss (the patients cannot measure their alcohol consumption) and/ or dishonest behavior. Therefore, this research proposes to develop an objective and quantitative method for the detection of alcohol addiction.
Challenges: As alcohol addiction results in changes in brain dynamics, hence, it is vital to investigate and develop a method based on brain activity. However, the main challenge in developing such an objective and quantitative method lies in its implementation for screening in smaller clinical setups. This limits the investigation to electroencephalogram (EEG) which is low cost, highly mobile and has good temporal resolution. Other modalities like MRI, PET etc are not feasible to be employed in smaller clinical settings.
Solution: With current innovations in brain EEG signals, the brain pathways involved in addiction can be investigated. In the last few decades, EEG research has been used to understand the complex underlying processes associated with the pathophysiology of addiction. Interpreting such processes using brain networks using EEG can not only help in diagnosing addiction but also assist in treating addiction. This research aims to develop neuromarker(s) based on brain network interpretation using EEG. The neuromarker will involve the features extraction and corresponding development of the machine learning model.
Few Words About Supervision: I have recently moved to FIT at Brno University of Technology. I have decade long experience of working in the field of neuro-signal and neuroimage processing and I am currently in the process of setting up a research group in this area. This is a multidisciplinary project and it will involve working with clinicians. However, the core of the project is related to IT in terms of development of a new method. Please feel free to contact me at malik@fit.vutbr.cz
Různé typy logik a automatů patří mezi nejzákladnější objekty studované a aplikované v oblasti informatiky již desítky let. Přesto v této oblasti existuje řada dosud neuspokojivě vyřešených problémů a neustále se objevují nové, vzrušující problémy související se stále novými aplikacemi logik a automatů (např. při formální verifikaci konečně i nekonečně stavových systémů s různými pokročilými řídicími či datovými strukturami, v rozhodovacích procedurách, při syntéze programů či hardware nebo i v metodách efektivního vyhledávání v různých typech dat či v síťovém provozu). Předmětem disertační práce bude primárně rozvoj současného stavu v oblasti efektivní práce s různými logikami (např. nad ukazatelovými strukturami, řetězci, různými aritmetikami, temporálními logikami apod.). Za tím účelem budou zkoumány přístupy založené na různých typech rozhodovacích diagramů, automatů, ale také např. přístupy založené na existenci modelu omezené velikosti či na efektivních redukcích mezi různými typy logických teorií. V souvislosti s tím pak budou rozvíjeny i metody efektivní práce s rozhodovacími diagramy a různými typy automatů (automaty nad slovy, stromy, nekonečnými slovy, automaty s čítači apod.). Práce zahrne jak teoretický výzkum, tak také prototypovou implementaci navržených technik a jejich experimentální vyhodnocení. Práce bude řešena ve spolupráci s týmem VeriFIT zabývajícím se na FIT VUT rozvojem technik pro práci s logikami a automaty a jejich aplikacemi. Jedná se zejména o dr. O. Lengála, jež bude působit v roli školitele specialisty, nebo také doc. L. Holíka či doc. A. Rogalewicze. V případě zodpovědného přístupu a kvalitních výsledků je zde možnost zapojení do grantových projektů (včetně mezinárodních). Je zde rovněž možnost úzké spolupráce s různými zahraničními partnery VeriFIT: Academia Sinica, Tchaj-wan (prof. Y.-F. Chen); TU Vienna, Rakousko (doc. F. Zuleger); LSV, ENS Paris-Saclay (prof. M. Sighireanu); IRIF, Paříž, Francie (prof. A. Bouajjani, doc. P. Habermehl, doc. C. Enea), Verimag, Grenoble, Francie (doc. R. Iosif); Uppsala University, Švédsko (prof. P.A. Abdulla, prof. B. Jonsson); či School of Informatics, University of Edinburgh, Velká Británie (prof. R. Mayr). V rámci tématu se student může také aktivně zapojit do různých grantových projektů, jako jsou např. projekty GA ČR GA23-07565S "ROULETTE - Reprezentace Booleovských funkcí pomocí adaptabilní datové struktury", GA23-06506S "AIDE - Pokročilá analýza a verifikace pro pokročilý software", ERC.CZ projekt LL1908 "Efektivní konečné automaty pro automatické usuzování" či některý z aktuálně připravovaných Horizon Europe projektů.
Tutor: Vojnar Tomáš, prof. Ing., Ph.D.
The topic focuses embedded image, video and/or signal processing. Its main goal is to research capabilities of "smart" and "small" units that have such features that allow for their applications requiring smyll, hidden, distributed, low power, mechanically or climatically stressed systems suitable of processing of some signal input. Exploitation of such systems is perspective and wide and also client/server and/or cloud systems. The units themselves can be based on CPU/DSP/GPU, programmable hardware, or their combination. Smart cameras can be considered as well. Applications of interest include:
A possibility exists in collaboration on grant projects, especially the newly submitted TAČR, H2020, ECSEL ones (potentially employment or scholarship possible).
In contrast to component-based dialogue systems (such as speech recognition, natural language generation, dialogue management and speech synthesis), this topic concentrates on end-to-end modelling of voice dialogues. The goal is to represent dialogues in a vector space, where it will be possible to semi-automatically or fully automatically design conversational models in the form of finite-state automata graphs from an unannotated set of voice or text training dialogues. This will lead to trustworthy and explainable modeling of conversations.
Speaker diarization (SD, determining who spoke when) is a crucial part of speech data mining and artificial intelligence (AI). It is crucial for down-stream algorithms, e. g. automatic speech recognition (ASR). Current SD performs well on many conditions but fails to handle overlapped speech. more than two speakers and realistic recordings (diverse acoustic conditions and speaking styles). Moreover, most current SD characterize speakers only using the acoustic information. Future SD will use an amalgam of inputs to enhance performance using all possible information resources, and this PhD topic proposes significant advances towards this goal. We will develop new architectures that extend the end-to-end SD paradigm to different multi-task scenarios. We also propose to integrate the processing of multi-stream inputs exploiting complementary information. The ultimate goal of the project is to combine all such systems into a unified framework that will substantially improve the performance of SD.
Cílem disertační práce je výzkum modelů vestavěné inteligence, která explicitně pracuje s energetickou náročností konkrétních operací a optimalizuje svoji činnost na základě konkrétních omezení na straně jednotlivých zařízení, případně celého systému. Součástí bude i realizace vybraných modelů na vhodném typu hardware, který bude možné využít v mezinárodních projektech, na jejichž řešení se vedoucí podílí.
Ukazuje se, že metody syntézy číslicových obvodů využívající evolučních algoritmů, zejména kartézského genetického programování pracujícího přímo nad reprezentací na úrovni hradel, jsou schopny produkovat implementace, které jsou v řadě případů mnohem efektivnější (typicky kompaktnější) nežli implementace získané pomocí současných syntézních technik využívajících interní reprezentace (např. AIG) a iterativní aplikace deterministických přepisovacích pravidel. Typickým cílem optimalizace je redukovat počet hradel optimalizovaného obvodu. V praxi se však vyskytuje požadavek optimalizovat obvod z hlediska více kriterií (např. zpoždění, plocha na čipu). V případě využití systému pro účely resyntézy je multikriteriální optimalizace nutností z důvodu zachování zpoždění obvodu, jehož část je předmětem optimalizace. Cílem disertační práce je navázat na předchozí výzkum a zabývat se možnostmi multikriteriální optimalizace číslicových obvodů s ohledem na dobrou škálovatelnost. Dále se předpokládá využití alternativních reprezentací jako je např. majority uzel, které lépe odrážejí principy nových technologií.Výzkum spadá do témat řešených výzkumnou skupinou Evolvable Hardware.
Tutor: Vašíček Zdeněk, doc. Ing., Ph.D.
Použití některých metod strojového učení, například v poslední době populárních hlubokých neuronových sítí, přináší problémy architektury tzv. černé skříňky, která sice může v některých případech správně rozhodovat, ale není možné snadno interpretovat způsob rozhodování, ověřovat, v jakém kontextu jsou závěry ještě věrohodné a nakolik mohou vést drobné změny vstupu ke zcela jiným závěrům.Cílem disertační práce je rozvinout existující přístupy k měření "dokazatelně správných" modelů umělých neuronových sítí a propojit je s technikami generování konfliktních (adversarial) příkladů, aby bylo možné kontrolovat a revidovat existující řešení, využívaná v praxi. Součástí práce bude i aplikace zkoumaných metod v rámci evropských projektů, na jejichž řešení se školitel podílí.
The project is concerned with advanced methods of image processing and generative AI. The aim is to research new methods using machine learning, in particular deep neural networks.
The project deals with image and video quality assessment metrics (IQM). The aim is to explore new ways how to incorporate human visual system properties into IQM. In particular, we will consider perception of HDR images, and utilization of additional knowledge (in form of metadata, 3D information, etc.) about the tested scenes using machine learning (e.g. neural networks).
The topic of identifying and extracting specific information from documents on the Web has been the subject of intensive research for quite a long time. The basic obstacles that make this problem difficult are the loose structure of HTML documents and absence of meta-information (annotations) useful for recognizing the content semantics. This missing information is therefore compensated by the analysis of various aspects of web documents that include especially the following:
A background knowledge about the target domain and the commonly used presentation patterns is also necessary for successful information extraction. This knowledge allows a more precise recognition of the individual information fields in the document body.
Current approaches to information extraction from web documents focus mainly modeling and analyzing the documents themselves; modeling the target information for more precise recognition has not yet been examined in detail in this context. The assumed goals of the dissertation are therefore the following:
Experimental implementation of the proposed methods using existing tools and experimental evaluation on real-world documents available on the WWW is also an integral part of the solution.
Tutor: Burget Radek, doc. Ing., Ph.D.
Topic description:
Today's computer systems and network elements record hundreds or thousands of events in log files that describe standard and non-standard device behavior or ongoing communications. By analyzing these events, it is possible to describe the typical behavior of a given device and detect anomalies caused, for example, by cyber-attacks.
Research includes the use of advanced machine learning and artificial intelligence techniques to detect anomalies based on log data. The topic includes designing a behavioral model for processing log events, representing events using features, and building a behavioral model based on the training data. Machine learning methods, time series or AI models can be used for anomaly detection.
The goal of the research is to propose efficient methods for automated analysis and anomaly detection of log information and to demonstrate how this method can be used to ensure the cybersecurity of computer systems.
Although there are technologies that allow publishing data on the WWW in machine-readable form (such as JSON-LD, RDFa, etc.), a large amount of structured data is still published on the web in the form of plain HTML/CSS code, which greatly limits the possibilities of their further use.Recently, new machine learning methods (especially deep learning methods) are gaining importance, which show interesting results, e.g., in recognizing important entities in weakly structured or unstructured data (e.g., text or images). However, the area of web document processing has not received much attention from this perspective. Existing works deal with the identification of simple data items and neglect structured data and more complex usage scenarios.The goal of this topic is to analyze and develop web content models suitable as input for machine learning and, at the same time, machine learning methods suitable for recognizing structured data in web documents.
V poslední době se stává čím dál patrnější, že k překlenutí propasti mezi současnými nejlepšími modely strojového učení a lidským učením nestačí jen zvyšovat počty parametrů a čekat na výkonnější hardware, který zvládne zpracování bilionů parametrů. Zdá se, že je třeba hledat nové modely, schopné objevovat a uvažovat na úrovni vysokoúrovňových pojmů a vztahů mezi nimi.Cílem disertační práce je výzkum nových modelů strojového učení, které překonají potřebu enormního množství příkladů, které jsou potřeba pro naučení chování, zvládnutelného lidmi velmi rychle (například počítač potřebuje sehrát velké množství her ke zvládnutí jednoduché videohry, zatímco člověk to zvládne velmi rychle, lidé ze sady proměnných rychle určí, jaká je příčinná souvislost mezi nimi, dokáží argumentovat sledem úvah atd.), a omezí problém sebejistého chybování (overconfident incorrectness) současných modelů. Budou zkoumány postupy učení, přidávající iterativně nové relevantní informace a také metody, podporující přímé pravděpodobnostní odvozování. Výsledky budou demonstrovány na vybraných problémech, zahrnujících mj. vysvětlování videa či tvorbu inferenčních grafů, operujících nad pojmy a vztahy mezi nimi.
The topic concerns algorithms of computer graphics and image synthesis. Its main goal is to research new algorithms so that their features and application possibilities are better understood so that they are improved or newly created. If suitable, it is possible to work on various platforms, includeing parallel CPUs, such as x86/64, ARM, Xeon PHI, GPU, etc. or other cores in CUDA, OpenCl, VHDL, etc. Algorithms of interest include:
Collaboration on grant projects, such as TACR, H2020, ECSEL possible (employment or scholarship).
Problem Statement: The importance of mental health has increased significantly over the past decade. However, the methods for the assessment of mental health issues at early stages are still in their infancy compared to the availability of corresponding methods for early assessment of physical health issues. Hence, it is required that due research is done to develop methods for early assessment of abnormalities leading to mental health problems.
Issues with Current Solutions: Unlike physical health parameters, the mental health is assessed through a number of subjective parameters. Hence, there is lack of objective and quantitative methods for mental health assessments. In addition, the patients seek help when their mental health problem is at advanced stage. So, there is lack of continuous monitoring for mental health issues.
Challenges: Many of the abnormalities related to mental health issues are subtle in nature and are related to behavior and other changes in facial expressions, speech and handwriting. In addition, there are changes in cortisol levels, skin conductance, heart rate variability and breathing rate. Hence, there are multiple modalities that should be included for measuring and quantifying any abnormalities related to mental health.
Solution: Every modality has its pros and cons. For example, in neuroimaging, functional magnetic resonance imaging has high spatial resolution (in mm) and low temporal resolution (in seconds) while electroencephalogram has low spatial resolution (in cm) and high temporal resolution (in milliseconds). Combining both of them will result in high spatial as well as high temporal resolution. This research deals with the assessment of abnormalities leading to mental health problems by utilizing multimodal approach. The various modalities may include, but not limited to, electroencephalogram (EEG) brain signals, facial videos, speech audios, handwriting and text from social media. The physiological parameters from various modalities include, but not limited to, the heart rate, breathing rate, dominant emotion, fatigue and stress. Dominant emotion can be classified as positive or negative and then sub-classified as sad, happy, angry etc. Data mining and data fusion techniques will be developed for this multimodal analysis. The corresponding multimodal data is available for this project.
Problem Statement: Stroke is a condition in which the supply of blood to the brain is restricted or stopped. When a stroke patient arrives at the hospital, a standard protocol is followed to provide the medical assistance to the patient as well as to assess the affect of the stroke on the brain. Generally, a second stroke may follow which can be devastating for the patient. Hence, it is critical to predict the next stroke and provide care that can avoid the next stroke or at least minimize its affects.
Issues with Current Solutions: The standard protocol at the hospital involves blood and urine tests as well as neuroimaging using CT and MRI scans. These tests are used to assess the damage done by the first stroke. However, the prediction of the next stroke depends on the doctor's experience and is very subjective. In many cases, the patient who is sent home after treatment, suffers the second more devastating stroke at home which can result in permanent disability.
Challenges: The standard protocol at the hospitals result in generation of lot of data, for example, hundreds of images from CT and MRI scans, hundreds of enzymes from blood and urine tests etc. The challenge is to collectively analyze all of this data and find correlations that can predict the second stroke for the patient.
Solution: This research will develop objective prediction method for occurrence of the second stroke from the data collected at the hospital using the standard protocol for stroke assessment, treatment and management. The analysis and development will involve machine learning techniques that handle multimodality data involving images, signals, text, and numbers.
Problem Statement: Among all the types of dementia, Alzheimer's disease (AD) is the most common form with 70 % of those affected by dementia having AD. As the prevalence of AD increases with age, the number of people living with AD is expected to rise over the next decades due to better quality of life that results in increase in age across many countries. All this has resulted in an increased focus on ensuring pre-onset detection of AD and the corresponding intervention, which can lead to slowing the progression of the disease by providing adequate diagnostics.
Issues with Current Solutions: Preclinical AD happens 10 to 15 years before the onset of the disease resulting in changes in the brain without showing any actual symptoms of the disease like memory loss etc. Pre-onset means detecting AD in or before the preclinical stage. The existing state-of-the-art methods mainly focus on the detection of later stages of AD, and the detection of preclinical AD is still an open research problem. Hence, this research targets pre-onset detection of AD (that is, early detection of Preclinical AD) because that will have huge impact on the lives of people. This can lead to early intervention and may result in further slowing the progression of the disease.
Challenges: At the stage of preclinical AD, the related signs and symptoms are not clear, and hence people at this stage do not seek any help. Therefore, a method for pre-onset detection of AD should be part of the regular health screening process and hence should be available in small clinical setups.
Solution: Method for detection of preclinical AD will involve investigating underlying brain mechanisms to monitor and track changes related to pre-onset detection of AD. Magnetic resonance imaging (MRI) will be used as a reference to investigate the brain dynamics however it cannot be used in practice due to its high-cost and specialized setup environment which limits its usage at the screening stage. Electroencephalogram (EEG) will be used in this research which is widely available, is low cost, has a good temporal resolution, and has high mobility. Therefore, this project aims to investigate the changes in underlying brain mechanisms using EEG to develop EEG-based neuromarker for pre-onset detection of AD. The neuromarker will involve the features extraction and corresponding development of the machine learning model.
In recent years, there has been a huge increase in the quality of the output of neural networks generating synthetic content, which has gone hand in hand with a significant simplification of the use of AI-based tools and their increased availability. Thus, the growing trend in the use of artificial intelligence brings new challenges to the field of cybersecurity. The most prominent examples are the use of "deepfakes" to attack biometric systems or the use of deep learning techniques to detect cyber attacks. The goal of this work is to analyze new trends, approaches, real attacks, their characteristics, impacts, and potential applications in a selected area of cybersecurity. The work should then propose new AI-based protection methods based on the analysis and research on the state of security for the selected areas.
Recommended areas of focus for the thesis:Impact of generative AI on code and application securityThe human factor in AI-based attacks - e.g., increasing the ability of humans to recognize these types of attacksImpacts of generative AI on the security of biometric authentication (voice, face, ...)
Zavedením sémantických operátorů do genetického programování umožnilo významně zefektivnit jeden ze základní stavebních pilířů evolučních výpočetních technik a zredukovat množství generací potřebných k nalezení řešení. Cílem disertační práce je navázat na předchozí výzkum a zabývat se možnostmi zavedení sémantických operátorů do kartézského genetického programování. Výzkum spadá do témat řešených výzkumnou skupinou Evolvable Hardware.
Research begins with the collection and analysis of indicators of compromise (IoC) from monitored network systems, which are then used to build a threat model of detected attacks. The model will then be shared and distributed to other networks to proactively improve the security of the network environment against detected threats.
The research includes the collection and analysis of threat data from anomaly detection systems. This data will be used to build a threat model using machine learning techniques or large language models (LLM). The created threat model will be distributed to other networks to proactively adapt the security of the networks against the detected threats.
Since the system shares sensitive data retrieved from detection systems, the solution requires ensuring the privacy of this data using federated learning.
Voice assistant-powered dialogue engines have previously been deployed in a number of commercial and governmental technological pipelines, with a diverse level of complexity. This PhD considers them as unstructured dialogues. The task is to comprehend those unstructured dialogues and translate them into explainable, safe, knowledge-grounded, trustworthy and bias-controlled machine learning models. Experiments will be conducted on laboratory data as well as in realistic scenarios provided by partners of the EU project Eloquence.
Tutor: Meduna Alexandr, prof. RNDr., CSc.
The growing trend in IT technology is increasing demands on users, who must make more and more decisions regarding IT security. As part of the solution of the thesis, there should be an introduction to security techniques and their usability. The thesis goal is to improve the usability of the selected security techniques to be effective in practice by concerning human factors knowledge and user-centered design principles. Primary interest of the work will be in the user perception of emerging technologies such as AI or changing trends in single and multi-factor authentication. Participation in relevant international conferences and publication in professional or scientific journals is expected.
The project deals with geo-localization in unknown environments using computer vision and computer graphics methods. The aim is to investigate and develop new image registration techniques (with geo-localized image database or 3D terrain model). The goal is an efficient implementation of proposed methods on mobile devices as well as search for additional applications in the area of image processing, computational photography, and augmented reality.