| Topic title |
Possible scientific supervisors |
Source of funding |
|
Artificial intelligence based method for predicting symptoms of early Dementia though the analysis of multimodal feature signals
|
prof. dr. Rytis MASKELIŪNAS |
state-funded |
Research Topic Summary.
A large proportion of our society is at risk of developing senile dementia, or at least being affected by some of the symptoms. Dementia is being diagnosed at younger and younger ages every year. The aim of this thesis is to address the challenge of early diagnosis of dementia, which is particularly relevant at an early stage of the disease, resulting in the greatest (if diagnosed timely) opportunity for individuals to delay the prognosis of the disease through a limited cycle of interventions and drug treatment. Challenges include the development and training of machine learning algorithms to predict symptoms in the early stages of dementia, based on a range of multimodal data such as EEG, ECG, eye pupils, fMRI, cognitive tests, etc.
|
|
Zero-day network intrusion recognition using artificial intelligence methods
|
prof. dr. Robertas DAMAŠEVIČIUS |
state-funded |
Research Topic Summary.
The continous development and extensive usage of the Internet benefits numerous networkusers. Meanwhile, network security becomes much more important with the wide usage of computer networks. The growing number of internet-connected systems makes them to become the targets of network attacks, resulting in serious damage. It is necessary to provide effective strategies to detect network attacks and maintain network secusity. Different kinds of cyberattacks need to be recognized in different ways. The increase in both the number and variety of new cyberattacks poses a tremendous challenge for cybersecurity systems that rely on a database of historical attack signatures. How to identify different kinds of network attacks thus becomes the main challenge in the domain of network cybersecurity to be solved, especially, the recognition of cyberattacks that have never been seen before (“zero-day attacks“). Various kinds of machine learning and deep learning methods can be used to classify network traffic and recognize network attacks without prior knowledge of their detailed characteristics. The aim of this dissertation is to develop, analyze and evaluate a method for the detection of “zero-day“ network attacks.
|
|
Deep neural network based monitoring methods of diabetes complications
|
prof. dr. Vidas RAUDONIS |
state-funded |
Research Topic Summary.
Relevance:
Global statistics show that 1 in 11 people, or just over 463 million people aged 20-79, have diabetes. Neuropathy develops during diabetes due to which nerve fibers in the whole body are damaged. One of the most complex complications of diabetes is diabetic foot syndrome, which is one of the leading causes of morbidity and mortality in diabetics. The treatment of diabetic foot requires professional and appropriate care, which depends on the complexity, type, extent of prevalence of the disease, and so on. The most common way out of an unattended and untreated diabetic foot is amputation of the lower limb. The patient is often unable to properly assess their own feet, for a variety of reasons, including inability to visually inspect them, a lack of competence in assessing the lesions that occur. The research will aim to address this issue by developing an automated monitoring system.
Scientific problem:
Complications of diabetes, specifically the diabetic foot, can occur in a variety of lower limb lesions, such as ulcers, wounds, nail lesions, limb deformities, and so on. Limb lesions are different, making accurate assessment of such lesions is difficult. Research aims to develop automated methods that can objectively assess the condition of the diabetic foot. However, it is not clear which telemetry and data processing and classification methods are most appropriate for monitoring diabetes complications.
Purpose:
Develop and investigate an artificial intelligence-based system for the timely monitoring of the diabetic foot.
Tasks:
1. Specification of monitoring system;
2. Development and expert evaluation of a representative database;
3. The development and research of a data processing and classification method;
4. Testing the system under laboratory conditions.
Expected results:
Dissemination of research results through participation in international conferences and publication in international journals. Closer cooperation between KTU and LSMU researchers in national and international projects.
|
|
Application of deep learning techniques for dynamic images anomality detection
|
doc. dr. Arūnas LIPNICKAS |
state-funded |
Research Topic Summary.
Recently emerged new artificial intelligence method called deep learning neural networks that has shown the ability to carry contextual analysis of video information. Such methods work perfectly with monolithic surfaces, but not jet applied in the food, textile and furniture industry with a patterned and mottled surfaces. Supervised approaches, however, can be challenging, since it is often difficult to obtain large amounts of labelled training data for abnormal class. The aim of this work is to build one-class unsupervised learning on fault-free samples by training a deep convolutional neural network to find dynamic images anomality.
|
|
Unsupervised, Deep Learning-Based Detection of Failures in Industrial Machinery
|
prof. dr. Rytis MASKELIŪNAS |
state-funded |
Research Topic Summary.
Despite spreading popularity, AI-based predictive maintenance is still not fully exploited in industrial plants worldwide. This is mainly because training predictive maintenance algorithms requires a large number of failure examples (known data), which is not always possible. For this reason, the aim of this thesis is to develop an efficient unsupervised predictive maintenance approach based on deep learning algorithmsfor the prediction and forecasting of failures in industrial machinery.
|
|
Application of Generative Adversarial Networks for novelty detection in one-class problems
|
doc. dr. Arūnas LIPNICKAS |
state-funded |
Research Topic Summary.
One-class novelty detection is a serious problem that receives relatively little attention in the debate about machine learning problems.
Generative Adversarial Networks (GAN) networks are finding more and more of their own applications because they need less input data for training compared to artificial-intelligence-based systems. The purpose of this work is to determine the capabilities of the GAN network to recognize the defect of the textured surface pattern in a limited number of samples of defect training.
|
|
Non - contact methods for monitoring of human vital signals
|
prof. dr. Vidas RAUDONIS |
state-funded |
Research Topic Summary.
Relevance:
The access to health care is limited and this is due to the COVID-19 pandemic. Older people have been hit hardest by this change, which has affected their physical and mental health, reduced their ability to function and made their illness worse. For this reason, there has been a great need for remote and contactless monitoring of human health. Research aims to address to this growing need and offer solutions that are appropriate for personal health care. The proposed solutions would ensure a safe and, above all, independent life for people who need constant health monitoring.
Scientific problem:
A person’s health status is initially assessed based on the person’s respiratory rate, heart rate, human body temperature, blood pressure, heart rate variation, and other vital signs. Mostly vital signals are measured by wearing specialized devices such as smart bracelets and watches. However, the use of these devices for monitoring vital signals is limited because older people in particular avoid wearing them or simply forget to put them on. It is therefore necessary to develop telemetry technologies that can accurately measure these signals in a non-contact manner. The development of such technologies requires the resolution of emerging scientific uncertainties. It is not clear what methods of computer vision can effectively and reliably detect the extremely small movements of the human body that occur when a person breathes. It is not clear what methods of artificial intelligence and optical measurement technologies can be used to measure blood pressure and with what accuracy.
Purpose:
To develop and investigate a system based on artificial intelligence methods for remote monitoring of human vital signals.
Tasks:
Specification of monitoring by determining the basic technical and software requirements of the system; Development and expert evaluation of a representative database; The development of a data processing and classification method is a study; Test of the system under laboratory conditions.
Expected results:
Dissemination of research results through participation in international conferences and publication in international journals.
|
|
Continuous bioprocess modeling and optimal control
|
prof. dr. Renaldas URNIEŽIUS |
state-funded |
Research Topic Summary.
Biotechnological processes are among the most complicated control objects due to nonlinear
relationships between the process variables, changing of process dynamic parameters in wide
ranges during fed-batch cultivation processes, lack in reliable sensors for direct measurement of
important state variables. Therefore, development of effective soft sensor systems is a relevant
bio-engineering task. Demand for such systems is high, they are relevant for development of new
and improvement of existing bio-processes both in research laboratories and industrial
companies.
The developed biotechnological process monitoring and optimal control algorithms and systems
would represent new results in bio-process state estimation and control practice. New knowledge
in the soft sensors development and applications will be obtained and the software technologies
for realization of the soft sensor systems will be developed and tested.
KTU Dept. of Automation is currently performing a scientific research grant “Development and
application of soft sensor system for state estimation of biotechnological processes” (No. 01.2.2-
LMT-K-718-03-0039), under which there are opportunities for new researchers to simultaneously
perform their research and participate in the compensated application development for biopharmaceutical
industry.
|
|
A study of Virtual Reality application for learning systems
|
prof. dr. Tomas BLAŽAUSKAS |
state-funded |
Research Topic Summary.
A large proportion of people have a dominant kinesthetic sensory preference. Besides, human learning depends on the environment around them. The emergence of high-quality and relatively inexpensive virtual reality devices, such as HTC Vive, Oculus Rift, Oculus Go, etc., has made it possible to realistically simulate the surrounding environment and simulate real-world activities, thus opening up opportunities for kinesthetic distance learning. The objectives of the research are: to analyze the methods of using virtual reality for learning; to develop the concept of a learning system that incorporates virtual reality; to propose a model of a multimodal system; to offer ways of providing lectures in virtual reality and ways to learn by using it; evaluate the effectiveness of using virtual reality for learning. The expected results are the proposed concepts, models, and practical implementations used for learning using virtual reality and the experimental evaluation of the proposed solutions.
|
|
Machine learning models for solving object shape recognition problems
|
doc. dr. Armantas OSTREIKA |
state-funded |
Research Topic Summary.
When applying artificial intelligence algorithms, it is important to properly select criteria that combine the essential properties of the object or phenomenon under consideration, and thus to create an appropriate criteria/parameter model that allows unambiguous identification and classification of object state changes. In computer vision tasks, the creation of such models is not a trivial task, as visual information is enormously diverse and, moreover, changes dynamically.
Properly designed and implemented, such models serve for high-level decision-making using a digital image or video data. These solutions are designed to automate tasks that can only be performed by a human. Computer vision systems are used in many fields, including manufacturing, medical, traffic monitoring, security systems, etc. Computer vision methods allow automatic analysis of important information in the processing of large data flows. Artificial intelligence algorithms in this case are an integral part of solving computer vision problems.
The proposed topic will aim to improve existing common methodologies by focusing on specific products and real-world processes for greater efficiency and reliability, adapting and improving machine learning methods in Decision trees, K nearest neighbors, Na?ve Bayes, Support vector classifiers, Direct propagation neural networks, and Convolutional neural networks.
Here, computer vision tasks will include methods for obtaining, processing, analyzing, and interpreting digital images; the extraction of multidimensional real-world data to obtain digital or symbolic information, for example, in the form of decision making. Interpreting data in this context means turning digital images into descriptions that can be linked to other action processes and suggest further appropriate solutions.
Focusing on real examples of provided services or production processes, the aim will be to apply and test the developed models and methodologies in practice. Examples of applications include computer vision systems that measure and count products or blanks, estimate their weight, shape, or volume, and inspect objects at high speed based on their predefined characteristics.
In this research, we expect to overcome the operative restrictions that hamper the adoption of new algorithms. It is expected to modify existing methodologies and implement novel research ideas based on artificial intelligence and new heuristic approaches.
|
| Cyber security artificial intelligence methods for predicting and managing energy consumption in mobile devices |
prof. dr. Jevgenijus TOLDINAS |
state-funded |
