Pavle Milošević
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Pavle Milošević has a total of 56 co-authors. Here are the top co-authors by number of shared publications:
- Bratislav Petrović
University of Belgrade
18 shared publications - Ana Poledica
University of Belgrade
8 shared publications - Aleksandar Rakićević
University of Belgrade
6 shared publications - Ivana Dragović
Faculty of Organizational Sciences, University of Belgrade, Jove Ilica 154, 11000 Belgrade, Serbia
5 shared publications - Milica Zukanović
Faculty of Organizational Sciences, University of Belgrade, Jove Ilića 154, Belgrade, 11000, Serbia
5 shared publications - Dragan Radojević
Mihajlo Pupin Institute
4 shared publications
Latest Publications
- Open Access
Classifying Expected Benefits of Collective Organization: A Rule-Based Approach for Non-Wood Forest Product Collectors in Serbia
Although heterogeneity among non-wood forest product (NWFP) collectors is well established, less is known about how collectors’ expected benefits from association can be translated into operational support profiles. This study aims to operationalize NWFP collectors’ expected benefits from association into support-priority groups and to assess whether these groups differ in collectors’ stated willingness to participate in establishing an association, with implications for planning institutional support in Serbia. Data were collected through a 2022 survey of 639 NWFP-collecting households, of which 465 households with complete service-item data and at least one named service formed the primary analytical sample. Ten services were grouped into three indices (administrative, information and education, and market and infrastructure) and classified using a complete 27-rule framework implemented through fuzzy inference. The framework generated low (14.8%), medium (54.4%), and high (30.8%) support-priority groups. The medium support-priority group was characterized by a mixed pattern of expected benefits, while the high support-priority group reflected stronger expectations across complementary support domains. Stated willingness to participate in establishing an association, which was not used to construct the classification, was reported by 68.2% of the low, 73.0% of the medium, and 90.7% of the high support-priority group. The overall association was statistically significant (χ2(2) = 19.03, p = 7.38 × 10−5; Cramer’s V = 0.21), with higher adjusted odds of stated willingness in the high group than in either other group, whereas the low–medium contrast was not statistically significant. In a sensitivity analysis (N = 615), assigning 150 households with no recorded service selections to Low under a zero-score assumption reduced willingness in that group to 25.6%. Fuzzy and crisp implementations classified all 80 possible input profiles identically; the rule-based groups were not more strongly associated with willingness than alternative classifications. These findings show how expected association benefits can be translated into interpretable model-generated groups that may serve as exploratory input for planning differentiated association-based institutional support in Serbia.
Open AccessClassifying Expected Benefits of Collective Organization: A Rule-Based Approach for Non-Wood Forest Product Collectors in Serbia
Although heterogeneity among non-wood forest product (NWFP) collectors is well established, less is known about how collectors’ expected benefits from association can be translated into operational support profiles. This study aims to operationalize NWFP collectors’ expected benefits from association into support-priority groups and to assess whether these groups differ in collectors’ stated willingness to participate in establishing an association, with implications for planning institutional support in Serbia. Data were collected through a 2022 survey of 639 NWFP-collecting households, of which 465 households with complete service-item data and at least one named service formed the primary analytical sample. Ten services were grouped into three indices (administrative, information and education, and market and infrastructure) and classified using a complete 27-rule framework implemented through fuzzy inference. The framework generated low (14.8%), medium (54.4%), and high (30.8%) support-priority groups. The medium support-priority group was characterized by a mixed pattern of expected benefits, while the high support-priority group reflected stronger expectations across complementary support domains. Stated willingness to participate in establishing an association, which was not used to construct the classification, was reported by 68.2% of the low, 73.0% of the medium, and 90.7% of the high support-priority group. The overall association was statistically significant (χ2(2) = 19.03, p = 7.38 × 10−5; Cramer’s V = 0.21), with higher adjusted odds of stated willingness in the high group than in either other group, whereas the low–medium contrast was not statistically significant. In a sensitivity analysis (N = 615), assigning 150 households with no recorded service selections to Low under a zero-score assumption reduced willingness in that group to 25.6%. Fuzzy and crisp implementations classified all 80 possible input profiles identically; the rule-based groups were not more strongly associated with willingness than alternative classifications. These findings show how expected association benefits can be translated into interpretable model-generated groups that may serve as exploratory input for planning differentiated association-based institutional support in Serbia.
- Open Access
- 27 August 2026
- Published by SAGE Publications in Journal of Sports Analytics
- Vol. 12
From rookie to contract extension: Modeling NBA player career trajectories with machine learning
This paper investigates machine learning-based prediction of players’ long-term career trajectories at different career stages. Specifically, we predict whether a player is likely to achieve Elite Player , High-Impact Starter or Rotation Player status based on data from their first one, three, and five NBA seasons. The target classes are expert-defined, subjective categories intended to capture meaningful differences in players’ career success and on-court impact. We collected, analyzed, and preprocessed a dataset of 942 players, described by 60 attributes representing their performance metrics and physical characteristics. We compare the classification performance of Multilayer Perceptron, Random Forest, and XGBoost, along with Logistic Regression as a benchmark. Furthermore, we conducted two experiments: one that takes into account the effect of NBA era player was active and another that neglects it. Finally, we identify the most important attributes contributing to classification outcome. Random Forest achieves the best performance after the rookie season (ROCAUC = 0.671 ± 0.039), indicating that career outcomes can be predicted only to a limited extent from rookie-season data. Performance improves substantially after three seasons (0.793 ± 0.025) and five seasons (0.859 ± 0.023), suggesting that additional performance history provides much stronger signals of long-term career success. PER is identified as the most important performance indicator.
- 27 August 2026
- Published by SAGE Publications in Journal of Sports Analytics
- Vol. 12
Open AccessFrom rookie to contract extension: Modeling NBA player career trajectories with machine learning
This paper investigates machine learning-based prediction of players’ long-term career trajectories at different career stages. Specifically, we predict whether a player is likely to achieve Elite Player , High-Impact Starter or Rotation Player status based on data from their first one, three, and five NBA seasons. The target classes are expert-defined, subjective categories intended to capture meaningful differences in players’ career success and on-court impact. We collected, analyzed, and preprocessed a dataset of 942 players, described by 60 attributes representing their performance metrics and physical characteristics. We compare the classification performance of Multilayer Perceptron, Random Forest, and XGBoost, along with Logistic Regression as a benchmark. Furthermore, we conducted two experiments: one that takes into account the effect of NBA era player was active and another that neglects it. Finally, we identify the most important attributes contributing to classification outcome. Random Forest achieves the best performance after the rookie season (ROCAUC = 0.671 ± 0.039), indicating that career outcomes can be predicted only to a limited extent from rookie-season data. Performance improves substantially after three seasons (0.793 ± 0.025) and five seasons (0.859 ± 0.023), suggesting that additional performance history provides much stronger signals of long-term career success. PER is identified as the most important performance indicator.
- 15 October 2025
- Published by Polish Information Processing Society PTI in Annals of Computer Science and Information Systems
- Vol. 43, 411-416
Evaluating Effectiveness of Nonlinear Dimensionality Reduction in Hedge Funds’ Returns Forecasting
- 15 October 2025
- Published by Polish Information Processing Society PTI in Annals of Computer Science and Information Systems
- Vol. 43, 411-416
Evaluating Effectiveness of Nonlinear Dimensionality Reduction in Hedge Funds’ Returns Forecasting
- 18 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 272-277
Student Perception of Using Artificial Intelligence in Education
Rapid artificial intelligence (AI) penetration into university education presents opportunities and challenges. One of the issues is how the study materials and general support correlate with the student's perceptions of AI. Teachers should also know in which aspects (or applications fields) of AI students are uneasy and feel fear or danger. In this regard, a survey has been conducted in three countries, Iceland, Slovakia, and Serbia, covering three different European regions and different but intertwined study programs. Our results in the survey opinion mining are intriguing in several ways. For example, it has been shown that business students are more likely to support mandatory AI courses than business informatics students. Next, students feel the danger of applying AI in medicine, even though AI is usually learning from high-quality medical records. In this work, we explored the statistically significant differences in responses, proposed recommendations, and identified topics for future surveys. For example, why are the recognized application fields problematic from the students’ perspective? And handling hesitation in answers because, in a rapidly developing field of AI, it is difficult to express opinions clearly.
- 18 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 272-277
Student Perception of Using Artificial Intelligence in Education
Rapid artificial intelligence (AI) penetration into university education presents opportunities and challenges. One of the issues is how the study materials and general support correlate with the student's perceptions of AI. Teachers should also know in which aspects (or applications fields) of AI students are uneasy and feel fear or danger. In this regard, a survey has been conducted in three countries, Iceland, Slovakia, and Serbia, covering three different European regions and different but intertwined study programs. Our results in the survey opinion mining are intriguing in several ways. For example, it has been shown that business students are more likely to support mandatory AI courses than business informatics students. Next, students feel the danger of applying AI in medicine, even though AI is usually learning from high-quality medical records. In this work, we explored the statistically significant differences in responses, proposed recommendations, and identified topics for future surveys. For example, why are the recognized application fields problematic from the students’ perspective? And handling hesitation in answers because, in a rapidly developing field of AI, it is difficult to express opinions clearly.
- Top Cited PapersOpen Access
- 10 October 2024
- Published by Springer Nature in Artificial Intelligence Review
- Vol. 57 (12) , 1-28
A systematic review of computer vision-based personal protective equipment compliance in industry practice: advancements, challenges and future directions
Computerized compliance of Personal Protective Equipment (PPE) is an emerging topic in academic literature that aims to enhance workplace safety through the automation of compliance and prevention of PPE misuse (which currently relies on manual employee supervision and reporting). Although trends in the scientific literature indicate a high potential for solving the compliance problem by employing computer vision (CV) techniques, the practice has revealed a series of barriers that limit their wider applications. This article aims to contribute to the advancement of CV-based PPE compliance by providing a comparative review of high-level approaches, algorithms, datasets, and technologies used in the literature. The systematic review highlights industry-specific challenges, environmental variations, and computational costs related to the real-time management of PPE compliance. The issues of employee identification and identity management are also discussed, along with ethical and cybersecurity concerns. Through the concept of CV-based PPE Compliance 4.0, which encapsulates PPE, human, and company spatio-temporal variabilities, this study provides guidelines for future research directions for addressing the identified barriers. The further advancements and adoption of CV-based solutions for PPE compliance will require simultaneously addressing human identification, pose estimation, object recognition and tracking, necessitating the development of corresponding public datasets.
- 10 October 2024
- Published by Springer Nature in Artificial Intelligence Review
- Vol. 57 (12) , 1-28
Top Cited PapersOpen AccessA systematic review of computer vision-based personal protective equipment compliance in industry practice: advancements, challenges and future directions
Computerized compliance of Personal Protective Equipment (PPE) is an emerging topic in academic literature that aims to enhance workplace safety through the automation of compliance and prevention of PPE misuse (which currently relies on manual employee supervision and reporting). Although trends in the scientific literature indicate a high potential for solving the compliance problem by employing computer vision (CV) techniques, the practice has revealed a series of barriers that limit their wider applications. This article aims to contribute to the advancement of CV-based PPE compliance by providing a comparative review of high-level approaches, algorithms, datasets, and technologies used in the literature. The systematic review highlights industry-specific challenges, environmental variations, and computational costs related to the real-time management of PPE compliance. The issues of employee identification and identity management are also discussed, along with ethical and cybersecurity concerns. Through the concept of CV-based PPE Compliance 4.0, which encapsulates PPE, human, and company spatio-temporal variabilities, this study provides guidelines for future research directions for addressing the identified barriers. The further advancements and adoption of CV-based solutions for PPE compliance will require simultaneously addressing human identification, pose estimation, object recognition and tracking, necessitating the development of corresponding public datasets.
- Open Access
- 9 September 2024
- Published by International Conference on Information Systems Development
Logic-Based Evaluation of Production Scheduling Rules Using Interpolative Boolean Algebra
This paper proposes a logic-based approach based on Interpolative Boolean Algebra (IBA) for multi-criteria evaluation of different priority and dispatching rules for production scheduling. Scheduling is crucial in optimizing operational activities, enabling efficient resource allocation within specific time constraints. While standard approaches to multi-criteria evaluation often use the weighted sum or weighted product method, they cannot capture logical and statistical relationships from the data. To address these limitations, we propose logical aggregation (LA) based on IBA, ensuring transparency and explainability in data aggregation. This paper evaluates the performance of six well-known priority and dispatching rules on 30 common benchmark instances of the job shop problem based on four scheduling criteria functions as input attributes. Analysis shows that the Critical Ratio rule performs the best, with Earliest Due Date also being a solid recommendation. This is a valuable insight for production managers unable to perform time-consuming simulations when facing tight deadlines.
- 9 September 2024
- Published by International Conference on Information Systems Development
Open AccessLogic-Based Evaluation of Production Scheduling Rules Using Interpolative Boolean Algebra
This paper proposes a logic-based approach based on Interpolative Boolean Algebra (IBA) for multi-criteria evaluation of different priority and dispatching rules for production scheduling. Scheduling is crucial in optimizing operational activities, enabling efficient resource allocation within specific time constraints. While standard approaches to multi-criteria evaluation often use the weighted sum or weighted product method, they cannot capture logical and statistical relationships from the data. To address these limitations, we propose logical aggregation (LA) based on IBA, ensuring transparency and explainability in data aggregation. This paper evaluates the performance of six well-known priority and dispatching rules on 30 common benchmark instances of the job shop problem based on four scheduling criteria functions as input attributes. Analysis shows that the Critical Ratio rule performs the best, with Earliest Due Date also being a solid recommendation. This is a valuable insight for production managers unable to perform time-consuming simulations when facing tight deadlines.
- Open Access
- 23 March 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (7) , 950
IBA-VNS: A Logic-Based Machine Learning Algorithm and Its Application in Surgery
The interpretability and explainability of machine learning (ML) approaches play a key role in the trustworthiness of ML models in various applications. The objective of this paper is to incorporate a logic-based reasoning in the ML model that is not only accurate but also interpretable and easily applied. More precisely, we propose a hybrid IBA-VNS approach based on interpolative Boolean algebra (IBA) and variable neighborhood search (VNS). IBA is chosen over traditional multi-valued and/or fuzzy logic techniques due to its consistency in preserving all Boolean axioms. The VNS heuristic is used for model training, i.e., determining the optimal logical aggregation function within the IBA framework for solving observed prediction problems. Obtained logic aggregation functions are easy to understand and may provide additional insight to the decision-maker. The proposed approach does not require any domain knowledge and is applicable in various domains. IBA-VNS is evaluated on several standard datasets. Further, IBA-VNS is applied to the real-world problem of predicting hospital length of stay (LOS), showing exceptional results in terms of interpretability and accuracy. In fact, the dataset is collected from the LabSerb program regarding colorectal surgeries in the period 2015–2023. The proposed approach extracted knowledge regarding the problem, i.e., the causal relations between the patient’s health condition and LOS, along with achieving an MAE of 1.144 days.
- 23 March 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (7) , 950
Open AccessIBA-VNS: A Logic-Based Machine Learning Algorithm and Its Application in Surgery
The interpretability and explainability of machine learning (ML) approaches play a key role in the trustworthiness of ML models in various applications. The objective of this paper is to incorporate a logic-based reasoning in the ML model that is not only accurate but also interpretable and easily applied. More precisely, we propose a hybrid IBA-VNS approach based on interpolative Boolean algebra (IBA) and variable neighborhood search (VNS). IBA is chosen over traditional multi-valued and/or fuzzy logic techniques due to its consistency in preserving all Boolean axioms. The VNS heuristic is used for model training, i.e., determining the optimal logical aggregation function within the IBA framework for solving observed prediction problems. Obtained logic aggregation functions are easy to understand and may provide additional insight to the decision-maker. The proposed approach does not require any domain knowledge and is applicable in various domains. IBA-VNS is evaluated on several standard datasets. Further, IBA-VNS is applied to the real-world problem of predicting hospital length of stay (LOS), showing exceptional results in terms of interpretability and accuracy. In fact, the dataset is collected from the LabSerb program regarding colorectal surgeries in the period 2015–2023. The proposed approach extracted knowledge regarding the problem, i.e., the causal relations between the patient’s health condition and LOS, along with achieving an MAE of 1.144 days.
- 12 November 2022
- Published by Springer Nature
- p. 80-95
An Approach to Corporate Credit Rating Prediction Using Computational Intelligence-Based Methods
- 12 November 2022
- Published by Springer Nature
- p. 80-95
An Approach to Corporate Credit Rating Prediction Using Computational Intelligence-Based Methods