GJoreski, Hristijan

Preferred name

GJoreski, Hristijan

Official Name

GJoreski, Hristijan

Main Affiliation

Web Site

https://scholar.google.com/citations?user=LOpq5BoAAAAJ&hl=en

Now showing 1 - 9 of 9

Some of the metrics are blocked by your
consent settings
Network Anomaly Detection using Federated Learning for the Internet of Things
(2022)
;
Jakimovski, Bojan
;
Pfitzner, Bjarne
;
;
Arnrich, Bert
The widespread use of IoT devices has contributed greatly to the continuous digitisation and modernisation of areas such as healthcare, facility management, transportation, and household. These devices allow for real-time mobile sensing, use input and then simplify and automate everyday tasks. However, like all other devices connected to a network, IoT devices are also subject to anomalous behaviour primarily due to security vulnerabilities or malfunction. Apart from this, they have limited resources and can hardly cope with such anomalies and attacks. Therefore, early detection of anomalies is of great importance for the proper functioning of the network and the protection of users’ personal data above all. In this paper, deep learning and federated learning algorithms are applied in order to detect anomalies in IoT network tra c. The results obtained show that all the models achieve high accuracy, with the FL models providing slight worse results compared to the DL models. However, with the increase in the amount of user data, the model based on federated learning is expected to have better results over time.
Some of the metrics are blocked by your
consent settings
OCOsense Glasses–Monitoring Facial Gestures and Expressions for Augmented Human-Computer Interaction: OCOsense Glasses for Monitoring Facial Gestures and Expressions
(2023-01)
;
Mavridou, Ifigeneia
;
Archer, James Archer William
;
Cleal, Andrew
;
Stankoski, Simon
Some of the metrics are blocked by your
consent settings
Differentially Private Federated Learningfor Anomaly Detection in eHealth Networks
(ACM, 2021-09-21)
;
Pfitzner, Bjarne
;
;
Rakovic, Valentin
;
Arnrich, Bert
Increasing number of ubiquitous devices are being used in the medical field to collect patient information. Those connected sensors can potentially be exploited by third parties who want to misuse personal information and compromise the security, which could ultimately result even in patient death. This paper addresses the security concerns in eHealth networks and suggests a new approach to dealing with anomalies. In particular we propose a concept for safe in-hospital learning from internet of health things (IoHT) device data while securing the network traffic with a collaboratively trained anomaly detection system using federated learning. That way, real time traffic anomaly detection is achieved, while maintaining collaboration between hospitals and keeping local data secure and private. Since not only the network metadata, but also the actual medical data is relevant to anomaly detection, we propose to use differential privacy (DP) for providing formal guarantees of the privacy spending accumulated during the federated learning.
Some of the metrics are blocked by your
consent settings
Machine Learning based Anomaly Detection in Ambient Assisted Living Environments
(2021-09)
;
Rakovic, Valentin
;
;
Pfitzner, Bjarne
;
Arnrich, Bert
Some of the metrics are blocked by your
consent settings
Machine learning model for air pollution prediction in Skopje, North Macedonia
(2020-07)
Andonovic, Viktor
;
;
;
;
The low quality of air, especially high concentration of particulate matter that have significant negative effect on human health and environment, is a global problem in urban areas. Thus, early air pollution prediction is an urgent need in Skopje, North Macedonia with highly increased concentration of particulate matter especially during the winter months. The objective of this paper is to develop machine learning model for predicting the air pollution in Skopje. The methods are based on processing the collected data from different measurement locations in Skopje, generating numerous weather and pollution features, and choosing the optimal parameters (hyperparameters) for the model. The information for the various pollutants were provided from the measurement stations located near the Faculty of Electrical Engineering and Information Technologies building. The measured data are gathered from the three sensor nodes that are collecting data for following parameters: particulate matter with 10 or less micrometres (PM10), particulate matter with 2.5 or less micrometres (PM2.5), CO and NO2, and sending these data to a server for online monitoring or off-line analysis. The pollution data, together with the weather information for temperature, humidity, wind speed, and wind direction were combined to train the prediction model. The results show that the weather information is correlated with the air pollution, which allows to train a model that predicts the air pollution based on the weather data and the historical data about the pollution. The experimental evaluation showed that the best performing model, XGBoost, achieves Mean Absolute Error for PM10 values of 6.8, 9.7, and 12.4 for the nodes 3, 2, and 1 respectively, and for PM2.5 values 6.36, 8.81 and 8 for nodes 3, 2 and 1 respectively.
Some of the metrics are blocked by your
consent settings
Comparing Classical Machine Learning and Deep Learning for Classification of Arrhythmia from ECG Signals
(2023-11)
Marija Bikova
;
Vesna Ojleska Latkoska
;
Some of the metrics are blocked by your
consent settings
Comparing Classical Machine Learning and Deep Learning for Classification of Arrhythmia from ECG Signals
(2023-11-30)
Bikova, Marija
;
;
Arrhythmia detection is a vital task for reducing the mortality rate of cardiovascular diseases. Electrocardiogram (ECG) is a simple and inexpensive tool that can provide valuable information about the heart’s electrical activity and detect arrhythmias. However, manual analysis of ECG signals can be time-consuming and prone to errors. Therefore, machine learning models have been proposed to automate the process and improve the accuracy and efficiency of arrhythmia detection. In this paper, we compare six machine learning models, namely ADA boosting, Gradient Boost, Random Forest, C-Support Vector (SVC), Convolutional Neural Network (CNN), and Long Short-Term Memory Network (LSTM), for arrhythmia detection using ECG data from the MIT-BIH Arrhythmia Database. We evaluate the performance of the models using various metrics, such as accuracy, precision, recall, and F1-score, on different classes of ECG beats. We also use confusion matrices to visualize the errors made by the models. We find that the CNN model is the best performing model overall, achieving accuracy of 95% and F1-score of 84.75%. SVC and LSTM were the second and third best, achieving accuracy of 94% and 93%, respectively. We also discuss the challenges of using ECG data for arrhythmia detection, such as noise, imbalance, and similarity of classes. We suggest some possible ways to overcome these challenges, such as using more advanced preprocessing and resampling techniques, or incorporating domain knowledge and expert feedback into the models.
Some of the metrics are blocked by your
consent settings
Deep Learning for Facial Expression and Human Activity Recognition Using Smart Glasses
(Institute of Electrical and Electronics Engineers (IEEE), 2025-03-14)
Marinova, Matea
;
Chona, Emilija
;
Kotevski, Andrej
;
Sazdov, Borjan
;
Kiprijanovska, Ivana
Some of the metrics are blocked by your
consent settings
Bias in vital signs? Machine learning models can learn patients’ race or ethnicity from the values of vital signs alone
(BMJ, 2025-07-10)
;
;
;
;
Mullan, Irene Dankwa
Objectives To investigate whether machine learning (ML) algorithms can learn racial or ethnic information from the vital signs alone. Methods A retrospective cohort study of critically ill patients between 2014 and 2015 from the multicentre eICU-CRD critical care database involving 335 intensive care units in 208 US hospitals, containing 200 859 admissions. We extracted 10 763 critical care admissions of patients aged 18 and over, alive during the first 24 hours after admission, with recorded race or ethnicity as well as at least two measurements of heart rate, oxygen saturation, respiratory rate and blood pressure. Pairs of subgroups were matched based on age, gender, admission diagnosis and disease severity. XGBoost, Random Forest and Logistic Regression algorithms were used to predict recorded race or ethnicity based on the values of vital signs. Results Models derived from only four vital signs can predict patients’ recorded race or ethnicity with an area under the curve (AUC) of 0.74 (±0.030) between White and Black patients, AUC of 0.74 (±0.030) between Hispanic and Black patients and AUC of 0.67 (±0.072) between Hispanic and White patients, even when controlling for known factors. There were very small, but statistically significant differences between heart rate, oxygen saturation and blood pressure, but not respiration rate and invasively measured oxygen saturation. Discussion ML algorithms can extract racial or ethnicity information from vital signs alone across diverse patient populations, even when controlling for known biases such as pulse oximetry variations and comorbidities. The model correctly classified the race or ethnicity in two out of three patients, indicating that this outcome is not random. Conclusion Vital signs embed racial information that can be learnt by ML algorithms, posing a significant risk to equitable clinical decision-making. Mitigating measures might be challenging, considering the fundamental role of vital signs in clinical decision-making.

GJoreski, Hristijan

Filters

Settings

Sort By

Results per page