Dominik Ozog
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Dominik Ozog has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Agnieszka Leśniak
Faculty of Civil Engineering, Cracow University of Technology, 31-155 Kraków, Poland
1 shared publication - Bartosz Leśniak
Faculty of Electrical Engineering, Automatics, Computer Science and Biomedical Engineering, AGH University of Krakow, 30-059 Kraków, Poland
1 shared publication - Dariusz Tymosiak
Faculty of Civil Engineering and Environmental Sciences, Bialystok University of Technology, 45E Wiejska Street, 15-351 Bialystok, Poland
1 shared publication - Ewa Błazik-Borowa
Faculty of Civil Engineering and Architecture, Lublin University of Technology, Nadbystrzycka 38D, 20-618 Lublin, Poland
1 shared publication - Maria Jolanta Sulewska
Department of Geotechnics, Roads and Geodesy, Faculty of Civil Engineering and Environmental Sciences, Bialystok University of Technology, Wiejska 45E Str., 15-351 Bialystok, Poland
1 shared publication - Marta Słowik
Lublin University of Technology
1 shared publication
Latest Publications
- Open Access
- 16 July 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (14) , 7150
Predicting Renovation Risk in Existing Buildings Using Multilayer Perceptrons: Correlation-Based Feature Screening and Model Architecture Comparison
Renovation projects in existing university buildings involve considerable uncertainty due to incomplete documentation, aging building systems, phased execution, and the need to maintain ongoing educational activities during construction. This study investigates multilayer perceptron (MLP) architectures as an exploratory proof-of-concept for mapping an expert-based renovation risk index expressed as a continuous value between 0 and 1. The analysis was based on 122 real renovation cases described by 13 input variables representing quantitative factors and encoded qualitative characteristics related to technical and organizational project conditions. Data preprocessing included qualitative data encoding and min–max normalization. The models were trained and evaluated using an 80/10/10 hold-out split with validation-based early stopping. Two MLP architectures developed in MATLAB R2025b were compared to assess the effect of network depth on predictive performance. Model performance was evaluated using the coefficient of determination (R2) and mean squared error (MSE). The model with two hidden layers achieved R2 ≈ 0.58 and MSE ≈ 0.065, whereas the model with four hidden layers achieved R2 ≈ 0.86 and MSE ≈ 0.010. An ordinary multiple linear regression model, fitted to the full dataset as a reference linear analysis, showed weak explanatory power (R2 = 0.151). The results suggest that, within this exploratory dataset and the adopted hold-out procedure, the deeper MLP architecture achieved a closer fit to the available data than the shallower architecture. The findings provide a feasibility-oriented contribution to risk-informed planning in public building renovation projects, while requiring confirmation through repeated resampling and external validation.
- 16 July 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (14) , 7150
Open AccessPredicting Renovation Risk in Existing Buildings Using Multilayer Perceptrons: Correlation-Based Feature Screening and Model Architecture Comparison
Renovation projects in existing university buildings involve considerable uncertainty due to incomplete documentation, aging building systems, phased execution, and the need to maintain ongoing educational activities during construction. This study investigates multilayer perceptron (MLP) architectures as an exploratory proof-of-concept for mapping an expert-based renovation risk index expressed as a continuous value between 0 and 1. The analysis was based on 122 real renovation cases described by 13 input variables representing quantitative factors and encoded qualitative characteristics related to technical and organizational project conditions. Data preprocessing included qualitative data encoding and min–max normalization. The models were trained and evaluated using an 80/10/10 hold-out split with validation-based early stopping. Two MLP architectures developed in MATLAB R2025b were compared to assess the effect of network depth on predictive performance. Model performance was evaluated using the coefficient of determination (R2) and mean squared error (MSE). The model with two hidden layers achieved R2 ≈ 0.58 and MSE ≈ 0.065, whereas the model with four hidden layers achieved R2 ≈ 0.86 and MSE ≈ 0.010. An ordinary multiple linear regression model, fitted to the full dataset as a reference linear analysis, showed weak explanatory power (R2 = 0.151). The results suggest that, within this exploratory dataset and the adopted hold-out procedure, the deeper MLP architecture achieved a closer fit to the available data than the shallower architecture. The findings provide a feasibility-oriented contribution to risk-informed planning in public building renovation projects, while requiring confirmation through repeated resampling and external validation.
- Open Access
Testing Using the DCP Probe of a Subgrade Modeled from Difficult-to-Compact Sand in a Calibration Chamber
The aim of the article is to analyze the possibilities of using a lightweight dynamic cone probe DCP to determine the quality of compaction of surface layers of embankments (from 0.10 m to approx. 0.80 m below ground level). For this purpose, comparative tests of non-cohesive soil used for the construction of embankments were carried out using the DCP test and direct tests of the degree of compaction IS in a calibration chamber with the following dimensions: height 1.10 m and diameter 0.75 m. The subsoil was prepared from difficult-to-compact sand (Sa) with a uniformity coefficient of CU = 3.10 and curvature coefficient of CC = 0.99. The soil in the laboratory in the calibration chamber was compacted in layers using a vibratory plate compactor. A database for statistical analysis was obtained, n = 68 cases described by seven variables: z, ρ, w, ρd, IS, PI, N10(DCP). It was found that the DCP probe can be used to assess the degree of compaction of embankments made of non-cohesive soil, using the developed relationship IS = f(z, N10(DCP)).
Open AccessTesting Using the DCP Probe of a Subgrade Modeled from Difficult-to-Compact Sand in a Calibration Chamber
The aim of the article is to analyze the possibilities of using a lightweight dynamic cone probe DCP to determine the quality of compaction of surface layers of embankments (from 0.10 m to approx. 0.80 m below ground level). For this purpose, comparative tests of non-cohesive soil used for the construction of embankments were carried out using the DCP test and direct tests of the degree of compaction IS in a calibration chamber with the following dimensions: height 1.10 m and diameter 0.75 m. The subsoil was prepared from difficult-to-compact sand (Sa) with a uniformity coefficient of CU = 3.10 and curvature coefficient of CC = 0.99. The soil in the laboratory in the calibration chamber was compacted in layers using a vibratory plate compactor. A database for statistical analysis was obtained, n = 68 cases described by seven variables: z, ρ, w, ρd, IS, PI, N10(DCP). It was found that the DCP probe can be used to assess the degree of compaction of embankments made of non-cohesive soil, using the developed relationship IS = f(z, N10(DCP)).
- Open Access
Supported Influence Mapping for Mobile Robot Pathfinding in Dynamic Indoor Environments
Pathfinding is the process of finding the lowest cost route between a pair of points in space. The aforementioned cost can be based on time, distance, the number of required turns, and other individual or complex criteria. Pathfinding in dynamic environments is a complex issue, which has a long history of academic interest. An environment is considered dynamic when its topology may change in real time, often due to human interference. Influence mapping is a solution originating from the field of video games, which was previously used to solve similar problems in virtual environments, but achieved mixed results in real-life scenarios. The purpose of this study was to find whether the algorithm could be used in real indoor environments when combined with information collected by remote sensors.
Open AccessSupported Influence Mapping for Mobile Robot Pathfinding in Dynamic Indoor Environments
Pathfinding is the process of finding the lowest cost route between a pair of points in space. The aforementioned cost can be based on time, distance, the number of required turns, and other individual or complex criteria. Pathfinding in dynamic environments is a complex issue, which has a long history of academic interest. An environment is considered dynamic when its topology may change in real time, often due to human interference. Influence mapping is a solution originating from the field of video games, which was previously used to solve similar problems in virtual environments, but achieved mixed results in real-life scenarios. The purpose of this study was to find whether the algorithm could be used in real indoor environments when combined with information collected by remote sensors.
- Open Access
- 19 October 2024
- Published by Wydawnictwo Naukowe Gabriel Borowski (WNGB) in Advances in Science and Technology Research Journal
- Vol. 19 (1) , 15-24
The kinematics of the manipulator with 1-DOF joints and controlled by elastic, inner ties
This document presents a mathematical model of forward and inverse kinematics for a manipulator composed of joints with one degree of freedom al-lowing only the rotation of a Cartesian coordinate system around one axis. The manipulator is controlled by means of internal ties running from the...
- 19 October 2024
- Published by Wydawnictwo Naukowe Gabriel Borowski (WNGB) in Advances in Science and Technology Research Journal
- Vol. 19 (1) , 15-24
Open AccessThe kinematics of the manipulator with 1-DOF joints and controlled by elastic, inner ties
This document presents a mathematical model of forward and inverse kinematics for a manipulator composed of joints with one degree of freedom al-lowing only the rotation of a Cartesian coordinate system around one axis. The manipulator is controlled by means of internal ties running from the...
- Open Access
Cross-Domain Indoor Visual Place Recognition for Mobile Robot via Generalization Using Style Augmentation
The article presents an algorithm for the multi-domain visual recognition of an indoor place. It is based on a convolutional neural network and style randomization. The authors proposed a scene classification mechanism and improved the performance of the models based on synthetic and real data from various domains. In the proposed dataset, a domain change was defined as a camera model change. A dataset of images collected from several rooms was used to show different scenarios, human actions, equipment changes, and lighting conditions. The proposed method was tested in a scene classification problem where multi-domain data were used. The basis was a transfer learning approach with an extension style applied to various combinations of source and target data. The focus was on improving the unknown domain score and multi-domain support. The results of the experiments were analyzed in the context of data collected on a humanoid robot. The article shows that the average score was the highest for the use of multi-domain data and data style enhancement. The method of obtaining average results for the proposed method reached the level of 92.08%. The result obtained by another research team was corrected.
Open AccessCross-Domain Indoor Visual Place Recognition for Mobile Robot via Generalization Using Style Augmentation
The article presents an algorithm for the multi-domain visual recognition of an indoor place. It is based on a convolutional neural network and style randomization. The authors proposed a scene classification mechanism and improved the performance of the models based on synthetic and real data from various domains. In the proposed dataset, a domain change was defined as a camera model change. A dataset of images collected from several rooms was used to show different scenarios, human actions, equipment changes, and lighting conditions. The proposed method was tested in a scene classification problem where multi-domain data were used. The basis was a transfer learning approach with an extension style applied to various combinations of source and target data. The focus was on improving the unknown domain score and multi-domain support. The results of the experiments were analyzed in the context of data collected on a humanoid robot. The article shows that the average score was the highest for the use of multi-domain data and data style enhancement. The method of obtaining average results for the proposed method reached the level of 92.08%. The result obtained by another research team was corrected.