Publications Archives - Page 7 of 8

19.02.2022.27.10.2022.

A Scientific Paper on Parking Occupancy Detection Using Deep Learning

A paper titled “Image-Based Parking Occupancy Detection Using Deep Learning and Faster R-CNN”, authored by Z. Scekic, S. Cakic, T. Popovic and A. Jakovljevic, was presented at the 26th International Conference on Information Technology, IEEE IT2022. The paper was presented by a young researcher Ms Zoja Scekic, Faculty of Applied Sciences, on 17 February 2022 in the paper presentation session at IT2022. The paper discussed the use of machine learning and HPC to develop prediction models that could be embedded into edge AI setting for Smart parking solutions The effort was supported by the EuroCC Monteengro team.

ABSTRACT – Smart city is one area with the growing use of Internet of Things and Artificial Intelligence. The concept of smart cities relies on making quality of life better, and solving important problems, such as global warming, public health, energy and resources. Smart parking management is one of the smart city use cases. This paper describes the use of deep learning algorithms to process images of parking lots and determine their current occupancy. The development of prediction models was done using PKLot dataset with 12417 images, Detectron2 software library, and Faster R-CNN algorithm. The resulting models can be integrated into parking space sensors and used for building smart parking solutions, and thus lead to more efficient use of space in urban areas, reduced traffic congestion, as well as reducing parking surfing to minimum.

The paper presented at the 26th IEEE IT2022 Conference

12.02.2022.27.10.2022.

Master Thesis: Ethics of Artificial Intelligence

Ms Jelena Tijanic, a master student at the University of Donja Gorica just defended her Master thesis titled “Ethics and Artificial Intelligence”. The thesis was done at the Master academic studies “Statistics” (EMOS) under supervision of her mentor prof. dr Milica Vukotic.

ABSTRACT – Artificial intelligence is ubiquitous and enables many of our daily routines – booking flights, driving without a driver, supports decision-making in governments and the private sector. Artificial intelligence technology delivers outstanding results in highly specialized areas such as cancer screening and building an inclusive environment for people with disabilities. They also help combat global problems such as climate change and world hunger, and help reduce poverty by optimizing economic aid. But technology also brings unprecedented new challenges. We see increased gender and ethnic bias, significant threats to privacy, dignity, dangers of mass surveillance, and increased use of unreliable law enforcement technologies.

The first part of the thesis presents the basic problems that the world is facing and why the development of artificial intelligence is a potential threat to the future of mankind. A new recommendation adopted by UNESCO member countries was presented. The second part of the thesis describes the basic concepts related to recommendation systems, the way they work, as well as their division. Here are some examples of where systems are used. Ethical problems that can be encountered during their development are described.Finally, a practical example of a movie recommendation system is described. The process of making the systems was described and the result analyzed.

Master Thesis: Ethics of Artificial Intelligence – Ms Jelena Tijanic

29.01.2022.27.10.2022.

AI for the Energy Sector: Forecasting Day-Ahead Electricity Metrics with Machine Learning

The day-ahead energy market lets market participants commit to buy or sell wholesale electricity one day before the operating day, to help avoid price volatility. Forecasting day-ahead electricity prices and loads creates basis for decision making in this process. Mr. Milutin Pavićević, a young researcher from the University of Donja Gorica , explored the possibility to utilize artificial neural networks in order to improve the forecasting day-ahead electricity prices and loads based on the historical data. This was the topic of his Master thesis research work done under supervision of professor Tomo Popovic, which finally resulted in a scientific article published in MDPI journal Sensors. The paper is titled ”Forecasting Day-Ahead Electricity Metrics with Artificial Neural Networks” within the Special Issue Complex Data Processing Systems and Computing Algorithms: New Concepts and Applications.

During this research effort the researchers engaged the domain experts which provided us with generous help in obtaining datasets and understanding the problem of day-ahead consumption, spot price prediction, and the electricity market. The results show the promising efficiency of AI and machine learning for the task of short-term prediction of electricity metrics. With the support of EuroCC Montenegro, the future work will include experimenting on the HPC infrastructure and creation of an industry pilot demonstration for the energy sector.

Forecasting Day-Ahead Electricity Metrics with Machine Learning

ABSTRACT – As artificial neural network architectures grow increasingly more efficient in time-series prediction tasks, their use for day-ahead electricity price and demand prediction, a task with very specific rules and highly volatile dataset values, grows more attractive. Without a standardized way to compare the efficiency of algorithms and methods for forecasting electricity metrics, it is hard to have a good sense of the strengths and weaknesses of each approach. In this paper, we create models in several neural network architectures for predicting the electricity price on the HUPX market and electricity load in Montenegro and compare them to multiple neural network models on the same basis (using the same dataset and metrics). The results show the promising efficiency of neural networks in general for the task of short-term prediction in the field, with methods combining fully connected layers and recurrent neural or temporal convolutional layers performing the best. The feature extraction power of convolutional layers shows very promising results and recommends the further exploration of temporal convolutional networks in the field.

The paper can be accessed at the Sensors website at the following link.

15.11.2021.27.10.2022.

UDG / EuroCC Montenegro will participate in the IEEE TELFOR 2021 conference

Researchers from the UDG / NCC Monteengro will be presenting the paper titled “Human Activity Detection Using Deep Learning and Bracelet with Bluetooth Transmitter”, authored by S. Cakic, S. Sandi, D. Nedic, S. Krco and T. Popovic, at the upcoming IEEE TELFOR 2021 conference. The conference will be taking place on 23-24 November 2022. The paper presents the use of AI/ML algorithms to implement human activity detection based on data collected from wearable IoT device and the research is done through collaboration with DNET Labs, an innovation technology company from Novi Sad, Serbia. More info on the conference is available at the following link.

ABSTRACT – The use of artificial intelligence, machine learning, and deep learning is finding its purpose in various fields nowadays. This paper describes a study in which Internet of
Things and deep learning are used to implement human activity detection based on data collected from bracelet equipped with Bluetooth transmitter. The main focus of the study was development of a prediction model using deep learning that would help elderly people and their caretakers. Time series data about elderly people activity was collected from bracelet using a Bluetooth gateway and IoT platform, and later annotated based on the activity logs they kept in a form of diary. A neural network is trained to classify data into two groups (binary classification problem) corresponding to activity of the person wearing the
bracelet. Initial study shows promising results of the presented approach for the use in human activity detection for elderly.

26.10.2021.27.10.2022.

Master thesis: Investigation of neural network efficiency in prediction electricity prices in the day-ahead market

Mr. Milutin Pavicevic just defended his Master thesis titled: “Investigation of neural network efficiency in prediction electricity prices in the day-ahead market”. The work focused on the use of artificial intelligence and exploration of various prediction models based on neural networks in order to improve prediction of electricity prices.

Mr. Pavicevic’s Master thesis defense at the Faculty for information systems and technologies, UDG

ABSTRACT – The power of neural networks in approximating continuous functions has led to more widespread use of this type of artificial intelligence in the field of time-series forecasting. This work examines the efficiency oftime-series prediction models when given the dataset of hourly values connected to day-ahead market of electrical energy. It presents the processing and windowing of the data to fit the prediction models, describes the specifics of the day-ahead market of electrical energy and more closely describes the way each of the used neural network models works.
The work looks at created neural network models with dense layers, convolutional neural networks (CNN) and recurrent neural networks (LSTM), and measures their performance. Testing results show their accuracy when predicting based on the dataset of hourly values of day-ahead electricity on the HUPX market, coupled with the hourly weather data, as well as the related dataset of the hourly values of electricity consumption in Montenegro.

Exploring the use of AI and ANNs for prediction of electricity prices in the day-ahead market

12.09.2021.27.10.2022.

UDG / EuroCC Montenegro will take part at the IEEE IDAACS 2021 conference

UDG researchers will be presenting the paper titled “Forecasting Day-Ahead Electricity Price with Artificial Neural Networks: a Comparison of Architectures”, authored by M. Pavicevic and T. Popovic, at the upcoming IEEE IDAACS 2021 conference. The conference will be taking place on 22-25 September 2021. The paper presents the use of AI/ML algorithms and CNN architectures for prediction of prices for the electric energy markets. More info on the conference is available at the following link.

ABSTRACT – The spot price prediction for the electric energy markets is a widely approached problem, used by many participants in the market. The ever-shifting rules and regulations, rising percentage of the electricity on the market being produced by solar and wind plants and many stochastic factors influencing it make the market price of electricity very volatile and hard to forecast. Many methods are used to tackle this problem, and their efficiency varies from dataset to dataset. In this work, we use the dataset of hourly day-ahead spot prices from the Hungarian HUPX market, and couple it with weather data for Hungary. We test various types of Dense, Recurrent and Convolutional neural network architectures and report on the results.

M. Pavicevic, a Master candidate at the UDG, will be presenting at the IEEE IDAACS 2021

29.08.2021.27.10.2022.

UDG researchers will be presenting a paper at IEEE IcETRAN 2021 Conference

Members of our EuroCC Montenegro team will be presenting the paper titled “Combined adaptive load balancing algorithm for parallel applications”, authored by L. Filipovic, B. Krstajic, and T. Popovic, at the upcoming IcETRAN 2021 conference. The conference will be taking place on Sep 8-10, 2021. More info on the conference is available at the following link.

ABSTRACT – Development and improvement of efficient techniques for parallel task scheduling on multiple cores processors is one of the key issues encountered in parallel and distributed computer systems. The purpose of process distribution improvement in parallel applications is in increased system performance, reduced application execution time, reduced losses and increased resource utilization.

This paper presents combined adaptive load balancing algorithm based on domain decomposition and master-slave
algorithms and its core scheduling adaptive mechanism that handles load redistribution according obtained and analyzed data. Selection of distribution algorithm, based on collected parameters and previously defined conditions, proved to deliver increased performances and reduced imbalance. Results of simulations confirm better performance of proposed algorithms compared to the standard algorithms reviewed in this paper.