ZNZoran Najdovski

Zoran Najdovski

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Zoran Najdovski has a total of 54 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    A comparative and critical study of EEGNet for fNIRS-driven cognitive load classification

    by Mehshan Khan, Houshyar Asadi, Li Zhang, Mohammad Reza Chalak Qazani, Ghazal Bargshady, Stefanos Gkikas, Christian Arzate, Sam Oladazimi, Zoran Najdovski, Lei Wei, et al.

    Accurately classifying cognitive load from functional near-infrared spectroscopy (fNIRS) signals remains a significant challenge due to temporal variability, inter-subject differences, and sensitivity to preprocessing choices. This study provides a comprehensive evaluation of EEGNet for fNIRS-based cognitive load classification by systematically examining the effects of temporal segmentation strategies (overlapping vs. non-overlapping), window lengths (10s, 20s, 30s), feature extraction methods (Analysis of Variance (ANOVA), Principal Component Analysis (PCA), Fast Independent Component Analysis (FastICA)), learning rate configurations (fixed and adaptive), and evaluation protocols (random split vs. subject-independent (SI)). Results from random-split experiments show that overlapping segmentation, combined with smaller fixed learning rates (0.01-0.001), yields the highest accuracies, due to temporal redundancy and dense sampling of hemodynamic transitions. However, SI evaluation reveals a substantial drop in accuracy, demonstrating limited generalization to unseen participants. Under SI evaluation, non-overlapping segmentation outperformed overlapping windows, with the best accuracy of 56.11% achieved using PCA features with a 20-second window and a 0.1 learning rate. These findings indicate that eliminating temporal redundancy helps the model learn more robust and generalizable representations of cognitive load across individuals. Although adaptive learning rate strategy improved training stability, it did not surpass the performance of optimally selected fixed learning rates. The study highlights the critical role of segmentation strategy and learning rate selection in improving model generalization and identifies methodological considerations essential for developing reliable, real-time, and SI cognitive load classification systems using fNIRS.

  • Open Access

    Machine Learning-Based Distraction Detection: A Systematic Review of Data, Algorithms, and Challenges

    by Mojtaba Mohammadi, Houshyar Asadi, Mohammadreza Chalak Qazani, Lei Wei, Zoran Najdovski, Hamid Khayyam, Chee Peng Lim

    Driver distraction is a major cause of road accidents globally. To address this, there has been significant research into using machine learning to monitor driver behavior. However, many existing reviews are limited in scope. They often focus only on specific algorithms or overlook the critical role of different data types. This paper fills that gap by providing an in-depth systematic review of both the data and the algorithms in machine learning-based driver distraction detection. By providing a clear overview of the current state of the field, this paper aims to improve the understanding of driver distraction detection and guide future research. To ensure a rigorous and unbiased selection process, the PRISMA methodology was used to select studies. To provide a fine-grained view, both topics (data and algorithms) are systematically categorized into specific subdomains. For each category, key concepts are explained, significant contributions are summarized, and their respective strengths and weaknesses are critically analyzed. In addition, this article provides a comparative and critical analysis of widely used driving-related datasets. Finally, challenges, research gaps, and possible directions for future research are analyzed and discussed. The main results of this review show that deep learning methods outperform older models, although they require more computational power. Issues such as model interpretability and monitoring systems with driving automation have received less attention. Also, systems that use multimodal data (end to end multimodal systems) have great potential for the future.

  • Open Access

    Optimized Deep Learning Framework for Emotion Recognition Using Multimodal Physiological Signals and Temporal Convolutional Networks

    by Mohsen Golafrouz, Houshyar Asadi, Mohammad Reza Chalak Qazani, Anwar Hosen, Zoran Najdovski, Lei Wei, Sam Oladazimi, Saeid Nahavandi

    Emotion recognition plays a crucial role in human–computer interaction, health monitoring, and affective computing by analysing physiological signals. Despite recent advancements, current research still faces challenges, including the lack of effective fusion strategies for diverse physiological modalities, difficulties in handling high-dimensional feature representations, and limited use of efficient temporal modelling techniques to capture complex emotional patterns. This study proposes a deep learning-based approach that fuses multiple physiological modalities, including Electroencephalography (EEG), Electrooculography (EOG), Electromyography (EMG), Galvanic Skin Response (GSR), Respiratory Rate (RR), Skin Temperature (SKT), and Photoplethysmography (PPG), to improve emotion recognition. Arousal and valence ratings were binarized into two classes (low/high) using a threshold of 4.5, formulating a binary classification problem. In addition to utilising Bidirectional Long Short-Term Memory (Bi-LSTM), the study employs Temporal Convolutional Networks (TCN), a widely used approach for time-series analysis, to efficiently capture temporal dependencies. The proposed model optimises feature selection through channel-wise strategies, incorporates advanced learning rate scheduling, and reduces computational overhead. Furthermore, window-wise, block-wise, and trial-wise evaluation protocols were investigated to assess the impact of temporal information leakage on emotion recognition performance. Using the DEAP dataset for validation, the proposed TCN-based approach achieved classification accuracies of 88.42% for valence and 86.35% for arousal under an overlapping block-wise evaluation protocol, demonstrating improved performance in binary emotion recognition and highlighting the importance of leakage-aware model assessment.


  • Open Access

    Automated Quantification of Fibrous Microplastics Using Attention Meta U-Net with Advanced Image Processing

    by Imran Hossain, Shofiqul Islam, Yi Zhang, Alessandra Sutti, Zoran Najdovski, Mohammad Anwar Hosen, Maryam Naebe

    The widespread release of microplastics (MPs), especially fibrous microplastics (FMPs) originating from synthetic textiles, poses a growing threat to environmental systems due to their persistence, mobility, and potential for bioaccumulation in aquatic and terrestrial ecosystems. Conventional gravimetric methods (GMs) remain the primary approach for assessing FMP shedding, yet they are hindered by moisture-sensitive filters, false positives from detergents and minerals, environmental contamination, and the labor-intensive manual measurement of individual fibers. To address these limitations, we developed an automated image analysis (AIA) framework that integrates an attention-based U-Net architecture with meta-learning modules to quantify FMP number, length, diameter, and mass from stitched microscopic images of entire filter membranes. This approach enables detection of fibers down to 28 μm in diameter with the spatial resolution of 2.17 µm/pixel, supports both target-color and multi-color analysis, and eliminates the need for manual characterization or extrapolation from partial membrane segments. The method achieved the highest accuracy of approximately 98% in color-specific fiber detection, correctly identifying 257 of 263 white fibers, and demonstrated similarly robust performance for black, red, and green fibers, while minimizing interference from non-target colors, even when their fibers overlapped. Multi-color detection was further validated using effluent water samples containing mixed-color fibers. Overall, the developed system enhances the accuracy, efficiency, and reproducibility of FMP analysis, offering a standardized and scalable approach for environmental monitoring of MP pollution.

  • Open Access

    Disturbed Delay-Aware Reinforcement Learning

    by Sathira Dilshan Bataduwaarachchi, Zoran Najdovski, Hieu Trinh, Van Thanh Huynh

    Reinforcement Learning (RL) is now one of the popular control methods used in the area of robotics, which extends to various robotic domains including applications with teleoperation. Considering such applications, it is inevitable that random communication delay exist in between agent-environment interactions. This delay inevitably causes latency in capturing observations and actions. Moreover, noise can occur through external disturbances in RL entities such as rewards which will result in perturbed values. Tackling complex environments in such a scenario can be challenging as the agent receives noisy measurements amid possible delays. We take measures to address this issue in delay-aware RL by proposing a novel reward correcting filter into the delay handling RL framework. The proposed algorithm will estimate the original reward using available perturbed reward values, smoothing out the noise values present in the reward, making an accurate reward estimate available to the agent. We theoretically prove how the proposed filter can estimate the reward with regards to the monotonicity and convergence of the delay-aware RL framework, which in turn will tackle complex environments like robotic tasks. The validation of the algorithm is carried out with experimentations using the DeepMind Control Suite, which highlighted significant performance by showcasing the new algorithm’s robustness with perturbed rewards under delay-aware RL frameworks.

  • Open Access

    Deterministic delay-aware reinforcement learning

    by Sathira Dilshan Bataduwaarachchi, Zoran Najdovski, Hieu Trinh, Chee Peng Lim, Van Thanh Huynh


  • Open Access

    Advancing Cognitive Load Detection in Simulated Driving Scenarios Through Deep Learning and fNIRS Data

    by Mehshan Ahmed Khan, Houshyar Asadi, Mohammad Reza Chalak Qazani, Ghazal Bargshady, Sam Oladazimi, Thuong Hoang, Ghazal Rahimzadeh, Zoran Najdovski, Lei Wei, Hirash Moradi, et al.

    The shift from manual to conditionally automated driving, supported by Advanced Driving Assistance Systems (ADASs), introduces challenges, particularly increased crash risks due to human factors like cognitive overload. Driving simulators provide a safe and controlled setting to study these human factors under complex conditions. This study leverages Functional Near-Infrared Spectroscopy (fNIRS) to dynamically assess cognitive load in a realistic driving simulator during a challenging night-time-rain scenario. Thirty-eight participants performed an auditory n-back task (0-, 1-, and 2-back) while driving, simulating multitasking demands. A sliding window approach was applied to the time-series fNIRS data to capture short-term fluctuations in brain activation. The data were analyzed using EEGNet, a deep learning model, with both overlapping and non-overlapping temporal segmentation strategies. Results revealed that classification performance is significantly influenced by the learning rate and windowing method. Notably, a learning rate of 0.001 yielded the highest performance, with 100% accuracy using overlapping windows and 97% accuracy with non-overlapping windows. These findings highlight the potential of combining fNIRS and deep learning for real-time cognitive load monitoring in simulated driving scenarios and demonstrate the importance of temporal modeling in physiological signal analysis.


  • Open Access

    Real Time Classification of Cognitive Load Using fNIRS and EEGNet in a Driving Simulation Task

    by Mehshan Ahmed Khan, Houshyar Asadi, Mohammad Reza Chalak Qazani, Ghazal Bargshady, Sam Oladazimi, Zoran Najdovski, Lei Wei, Hirash Moradi, Saeid Nahavandi

    Understanding and monitoring cognitive workload is essential for ensuring safety and performance in cognitively demanding environments such as driving. In this study, we investigate the effectiveness of Functional Near-Infrared Spectroscopy (fNIRS) in measuring and classifying cognitive workload during a realistic simulated driving task. Participants were required to perform a dual-task paradigm, which combined a primary driving task with a secondary auditory n-back task at three levels of difficulty: 0-back (low workload), 1-back (moderate workload), and 2-back (high workload). This multi-level workload design allowed us to capture a wide range of cognitive demands reflective of real-world multitasking scenarios. To analyze the hemodynamic responses recorded from the prefrontal cortex, we employed the EEGNet deep learning model, adapted for use with fNIRS signals. A Principal Component Analysis (PCA) feature selection method was applied to reduce dimensionality and isolate the most informative features, improving model performance and computational efficiency. We evaluated the model's performance using both overlapping and non-overlapping signal segmentation strategies across different window lengths (10s, 20s, and 30s). Our results show that a learning rate of 0.001 consistently produced the highest classification accuracy. Notably, a 30-second overlapping window achieved 100% accuracy, while 10-second non-overlapping segments yielded 97% accuracy, highlighting the influence of temporal segmentation on workload classification performance.

  • Open Access

    Advancing Cognitive Load Detection in Simulated Driving Scenarios Through Deep Learning and fNIRS Data

    by Mehshan Ahmed Khan, Houshyar Asadi, Mohammad Reza Chalak Qazani, Ghazal Bargshady, Sam Oladazimi, Thuong Hoang, Ghazal Rahimzadeh, Zoran Najdovski, Lei Wei, Hirash Moradi, et al.

    The shift from manual to conditionally automated driving, supported by Advanced Driving Assistance Systems (ADAS), introduces challenges, particularly increased crash risks due to human factors like cognitive overload. Driving simulators provide a safe and controlled setting to study these human factors under complex conditions. This study leverages Functional Near-Infrared Spectroscopy (fNIRS) to dynamically assess cognitive load in a realistic driving simulator during a challenging nighttime-rain scenario. Thirty-eight participants performed an auditory n-back task (0-, 1-, and 2-back) while driving, simulating multitasking demands. A sliding window approach was applied to the time-series fNIRS data to capture short-term fluctuations in brain activation. The data were analyzed using EEGNet, a deep learning model, with both overlapping and non-overlapping temporal segmentation strategies. Results revealed that classification performance is significantly influenced by learning rate and windowing method. Notably, a learning rate of 0.001 yielded the highest performance, with 100% accuracy using overlapping windows and 97% accuracy with non-overlapping windows. These findings highlight the potential of combining fNIRS and deep learning for real-time cognitive load monitoring in simulated driving scenarios and demonstrate the importance of temporal modeling in physiological signal analysis.


  • Open Access

    Advancing haptic realism: modelling grasp contact vibrations for enhanced virtual environment interaction

    by Zoran Najdovski, Siamak Pedrammehr, Mohammad Reza Chalak Qazani, Hamid Abdi, Houshyar Asadi

    In haptic technology, achieving realistic tactile feedback is crucial for enhancing a user's experience in virtual environments. Previous studies lack effective methods for transmitting high-frequency vibrations crucial for realistic tactile feedback in haptic interfaces, highlighting the need for our research to address this gap. This paper explores the application of a tactile gripping interface to transmit the high-frequency vibrations produced when contacting a hard object’s surface. These short vibrations improve the tactile sensation of hard virtual surfaces when overlaid on traditional position-based force feedback within a haptic environment. The enhanced realism of virtual objects is achieved by effectively estimating the vibration composition from user-induced parameters. This work presents a prototype grasping interface and empirically demonstrates this device’s utility. We examine empirical grasp contact data, recorded and interpreted, to recognise the relationship between dynamic user-controlled parameters and the resulting vibration transients. This relationship effectively incorporates these changing dynamics to model the grasp impact and estimate the essential system parameters to understand the influence of the user’s grasp force. Through our multi-point grasping interface design, this work demonstrates a mathematical relationship between the user’s grasp force and the high-frequency vibrations from contact with hard surfaces. The study found that the proposed haptic interface achieved an RMSE of 0.05, demonstrating a high level of accuracy. This low RMSE value signifies that the predicted vibrations closely matched the actual measured vibrations, validating the system’s capability to generate precise high-frequency transients. Such accuracy is critical for practical applications, including realistic tactile feedback in virtual environments, where precise modelling enhances user experience and interaction reliability. This work provides a foundational model for developing advanced haptic technologies, enabling more immersive virtual environments and precise control in teleoperation and training simulations.