Siamak Mehrkanoon
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Siamak Mehrkanoon has a total of 101 co-authors. Here are the top co-authors by number of shared publications:
- Johan A. K. Suykens
ESAT-STADIUS, KU Leuven
24 shared publications - Raghvendra Mall
Qatar Computing Research Institute
4 shared publications - Xiaolin Huang
Shanghai Jiao Tong University
4 shared publications - Danny Eric Paul Vanpoucke
Hasselt University Martelarenlaan 42
3 shared publications - Jie Shi
Utrecht University
3 shared publications - Katrien V. Bernaerts
Sustainable Polymer Synthesis Group, Aachen-Maastricht Institute for Biobased Materials (AMIBM), Faculty of Science and Engineering, Maastricht University, Urmonderbaan 22, 6167 RD Geleen, The Netherlands
3 shared publications
Latest Publications
- Open Access
- 17 October 2026
- Published by Springer Nature in BMC Bioinformatics
Genes2Reg: Constructing a biologically-faithful gene coupling prior via deep adaptive fusion of heterogeneous knowledge sources
- 17 October 2026
- Published by Springer Nature in BMC Bioinformatics
Open AccessGenes2Reg: Constructing a biologically-faithful gene coupling prior via deep adaptive fusion of heterogeneous knowledge sources
- Open Access
- 24 September 2026
- Published by Taylor & Francis in International Journal of Remote Sensing
- p. 1-31
Weakly supervised semantic segmentation for scalable and repeatable mapping of palsa landforms in rapidly thawing permafrost peatlands
1. Permafrost ecosystems, which act as a global carbon sink, are likely to become a significant source of carbon emissions as climate change causes permafrost to thaw and decrease in spatial extent...
- 24 September 2026
- Published by Taylor & Francis in International Journal of Remote Sensing
- p. 1-31
Open AccessWeakly supervised semantic segmentation for scalable and repeatable mapping of palsa landforms in rapidly thawing permafrost peatlands
1. Permafrost ecosystems, which act as a global carbon sink, are likely to become a significant source of carbon emissions as climate change causes permafrost to thaw and decrease in spatial extent...
- 15 September 2026
- Published by Springer Nature
- p. 641-653
Integrating Weather Station Data and Radar for Precipitation Nowcasting: SmaAt-fUsion and SmaAt-Krige-GNet
- 15 September 2026
- Published by Springer Nature
- p. 641-653
Integrating Weather Station Data and Radar for Precipitation Nowcasting: SmaAt-fUsion and SmaAt-Krige-GNet
- Open Access
- 1 July 2026
- Published by Springer Nature in Neuroinformatics
- Vol. 24 (3) , 52
Optimal Size of Electrocorticography Grids for Classification of Hand Movements
Implanted Brain-Computer Interfaces (BCIs) hold significant promise as a replacement for conventional assistive technologies for people with extensive motor impairments. In recent years, significant progress has been made in enhancing BCI performance, bringing clinically viable systems within reach. Some of these developments rely on increasingly large numbers of high spatial-density electrocorticography (ECoG) electrodes, which enable the extraction of spatially detailed information from extensive brain areas, but may also elevate surgical burden, posing a challenge for the clinical adoption of implanted BCIs. To mitigate this risk, we conducted an exhaustive investigation of hand movement classification performance involving 4, 5, and 8 classes in nine individuals with epilepsy, exploring all possible rectangular ECoG subgrids within the 32-, 64-, and 128-channel grids that were implanted in these individuals. Our findings reveal that the surface area of ECoG grids can be substantially reduced by 75–94% compared to the original grids, without a meaningful decline in classification performance, if electrodes are placed over informative areas. Classification performance across datasets was stable for progressively smaller subgrids until a critical threshold of approximately 60mm 2 was reached, below which performance declined substantially. We show that smallest subgrids with an area above the threshold achieved equally high classification F1 scores (range 81.64-99.71%) as the full grids with an area > 230mm 2 (range 82.85-96.75%). We conclude that ECoG-based BCI can be used to accurately decode up to seven different hand movements from well-located grids with a small number of electrodes, paving the way for smaller and safer BCI implants.
- 1 July 2026
- Published by Springer Nature in Neuroinformatics
- Vol. 24 (3) , 52
Open AccessOptimal Size of Electrocorticography Grids for Classification of Hand Movements
Implanted Brain-Computer Interfaces (BCIs) hold significant promise as a replacement for conventional assistive technologies for people with extensive motor impairments. In recent years, significant progress has been made in enhancing BCI performance, bringing clinically viable systems within reach. Some of these developments rely on increasingly large numbers of high spatial-density electrocorticography (ECoG) electrodes, which enable the extraction of spatially detailed information from extensive brain areas, but may also elevate surgical burden, posing a challenge for the clinical adoption of implanted BCIs. To mitigate this risk, we conducted an exhaustive investigation of hand movement classification performance involving 4, 5, and 8 classes in nine individuals with epilepsy, exploring all possible rectangular ECoG subgrids within the 32-, 64-, and 128-channel grids that were implanted in these individuals. Our findings reveal that the surface area of ECoG grids can be substantially reduced by 75–94% compared to the original grids, without a meaningful decline in classification performance, if electrodes are placed over informative areas. Classification performance across datasets was stable for progressively smaller subgrids until a critical threshold of approximately 60mm 2 was reached, below which performance declined substantially. We show that smallest subgrids with an area above the threshold achieved equally high classification F1 scores (range 81.64-99.71%) as the full grids with an area > 230mm 2 (range 82.85-96.75%). We conclude that ECoG-based BCI can be used to accurately decode up to seven different hand movements from well-located grids with a small number of electrodes, paving the way for smaller and safer BCI implants.
- Open Access
- 13 September 2025
- Published by MDPI AG in Algorithms
- Vol. 18 (9) , 580
Data-Efficient Sleep Staging with Synthetic Time Series Pretraining
Analyzing electroencephalographic (EEG) time series can be challenging, especially with deep neural networks, due to the large variability among human subjects and often small datasets. To address these challenges, various strategies, such as self-supervised learning, have been suggested, but they typically rely on extensive empirical datasets. Inspired by recent advances in computer vision, we propose a pretraining task termed “frequency pretraining” to pretrain a neural network for sleep staging by predicting the frequency content of randomly generated synthetic time series. Our experiments demonstrate that our method surpasses fully supervised learning in scenarios with limited data and few subjects, and matches its performance in regimes with many subjects. Furthermore, our results underline the relevance of frequency information for sleep stage scoring, while also demonstrating that deep neural networks utilize information beyond frequencies to enhance sleep staging performance, which is consistent with previous research. We anticipate that our approach will be advantageous across a broad spectrum of applications where EEG data is limited or derived from a small number of subjects, including the domain of brain-computer interfaces.
- 13 September 2025
- Published by MDPI AG in Algorithms
- Vol. 18 (9) , 580
Open AccessData-Efficient Sleep Staging with Synthetic Time Series Pretraining
Analyzing electroencephalographic (EEG) time series can be challenging, especially with deep neural networks, due to the large variability among human subjects and often small datasets. To address these challenges, various strategies, such as self-supervised learning, have been suggested, but they typically rely on extensive empirical datasets. Inspired by recent advances in computer vision, we propose a pretraining task termed “frequency pretraining” to pretrain a neural network for sleep staging by predicting the frequency content of randomly generated synthetic time series. Our experiments demonstrate that our method surpasses fully supervised learning in scenarios with limited data and few subjects, and matches its performance in regimes with many subjects. Furthermore, our results underline the relevance of frequency information for sleep stage scoring, while also demonstrating that deep neural networks utilize information beyond frequencies to enhance sleep staging performance, which is consistent with previous research. We anticipate that our approach will be advantageous across a broad spectrum of applications where EEG data is limited or derived from a small number of subjects, including the domain of brain-computer interfaces.
- Open Access
- 11 September 2025
- Published by Elsevier in Computers & Geosciences
- Vol. 207
Assessment of automated stratigraphic interpretations of boreholes with geology-informed metrics
- 11 September 2025
- Published by Elsevier in Computers & Geosciences
- Vol. 207
Open AccessAssessment of automated stratigraphic interpretations of boreholes with geology-informed metrics
- Open Access
Trans-XFed: An Explainable Federated Learning for Supply Chain Credit Assessment
This paper proposes a Trans-XFed architecture that combines federated learning with explainable AI techniques for supply chain credit assessment. The proposed model aims to address several key challenges, including privacy, information silos, class imbalance, non-identically and independently distributed (Non-IID) data, and model interpretability in supply chain credit assessment. We introduce a performance-based client selection strategy (PBCS) to tackle class imbalance and Non-IID problems. This strategy achieves faster convergence by selecting clients with higher local F1 scores. The FedProx architecture, enhanced with homomorphic encryption, is used as the core model, and further incorporates a transformer encoder. The transformer encoder block provides insights into the learned features. Additionally, we employ the integrated gradient explainable AI technique to offer insights into decision-making. We demonstrate the effectiveness of Trans-XFed through experimental evaluations on real-world supply chain datasets. The obtained results show its ability to deliver accurate credit assessments compared to several baselines, while maintaining transparency and privacy.
Open AccessTrans-XFed: An Explainable Federated Learning for Supply Chain Credit Assessment
This paper proposes a Trans-XFed architecture that combines federated learning with explainable AI techniques for supply chain credit assessment. The proposed model aims to address several key challenges, including privacy, information silos, class imbalance, non-identically and independently distributed (Non-IID) data, and model interpretability in supply chain credit assessment. We introduce a performance-based client selection strategy (PBCS) to tackle class imbalance and Non-IID problems. This strategy achieves faster convergence by selecting clients with higher local F1 scores. The FedProx architecture, enhanced with homomorphic encryption, is used as the core model, and further incorporates a transformer encoder. The transformer encoder block provides insights into the learned features. Additionally, we employ the integrated gradient explainable AI technique to offer insights into decision-making. We demonstrate the effectiveness of Trans-XFed through experimental evaluations on real-world supply chain datasets. The obtained results show its ability to deliver accurate credit assessments compared to several baselines, while maintaining transparency and privacy.
- Open Access
SSA-UNet: Advanced Precipitation Nowcasting via Channel Shuffling
Weather forecasting is essential for facilitating diverse socio-economic activity and environmental conservation initiatives. Deep learning techniques are increasingly being explored as complementary approaches to Numerical Weather Prediction (NWP) models, offering potential benefits such as reduced complexity and enhanced adaptability in specific applications. This work presents a novel design, Small Shuffled Attention UNet (SSA-UNet), which enhances SmaAt-UNet's architecture by including a shuffle channeling mechanism to optimize performance and diminish complexity. To assess its efficacy, this architecture and its reduced variant are examined and trained on two datasets: a Dutch precipitation dataset from 2016 to 2019, and a French cloud cover dataset containing radar images from 2017 to 2018. Three output configurations of the proposed architecture are evaluated, yielding outputs of 1, 6, and 12 precipitation maps, respectively. To better understand how this model operates and produces its predictions, a gradient-based approach called Grad-CAM is used to analyze the outputs generated. The analysis of heatmaps generated by Grad-CAM facilitated the identification of regions within the input maps that the model considers most informative for generating its predictions. The implementation of SSA-UNet can be found on our Github\footnote{\href{https://github.com/MarcoTurzi/SSA-UNet}{https://github.com/MarcoTurzi/SSA-UNet}}
Open AccessSSA-UNet: Advanced Precipitation Nowcasting via Channel Shuffling
Weather forecasting is essential for facilitating diverse socio-economic activity and environmental conservation initiatives. Deep learning techniques are increasingly being explored as complementary approaches to Numerical Weather Prediction (NWP) models, offering potential benefits such as reduced complexity and enhanced adaptability in specific applications. This work presents a novel design, Small Shuffled Attention UNet (SSA-UNet), which enhances SmaAt-UNet's architecture by including a shuffle channeling mechanism to optimize performance and diminish complexity. To assess its efficacy, this architecture and its reduced variant are examined and trained on two datasets: a Dutch precipitation dataset from 2016 to 2019, and a French cloud cover dataset containing radar images from 2017 to 2018. Three output configurations of the proposed architecture are evaluated, yielding outputs of 1, 6, and 12 precipitation maps, respectively. To better understand how this model operates and produces its predictions, a gradient-based approach called Grad-CAM is used to analyze the outputs generated. The analysis of heatmaps generated by Grad-CAM facilitated the identification of regions within the input maps that the model considers most informative for generating its predictions. The implementation of SSA-UNet can be found on our Github\footnote{\href{https://github.com/MarcoTurzi/SSA-UNet}{https://github.com/MarcoTurzi/SSA-UNet}}
- Open Access
- 23 April 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 282
TransCORALNet: A two-stream transformer CORAL networks for supply chain credit assessment cold start
- 23 April 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 282
Open AccessTransCORALNet: A two-stream transformer CORAL networks for supply chain credit assessment cold start
- Open Access
Self-supervised Spatial-Temporal Learner for Precipitation Nowcasting
Nowcasting, the short-term prediction of weather, is essential for making timely and weather-dependent decisions. Specifically, precipitation nowcasting aims to predict precipitation at a local level within a 6-hour time frame. This task can be framed as a spatial-temporal sequence forecasting problem, where deep learning methods have been particularly effective. However, despite advancements in self-supervised learning, most successful methods for nowcasting remain fully supervised. Self-supervised learning is advantageous for pretraining models to learn representations without requiring extensive labeled data. In this work, we leverage the benefits of self-supervised learning and integrate it with spatial-temporal learning to develop a novel model, SpaT-SparK. SpaT-SparK comprises a CNN-based encoder-decoder structure pretrained with a masked image modeling (MIM) task and a translation network that captures temporal relationships among past and future precipitation maps in downstream tasks. We conducted experiments on the NL-50 dataset to evaluate the performance of SpaT-SparK. The results demonstrate that SpaT-SparK outperforms existing baseline supervised models, such as SmaAt-UNet, providing more accurate nowcasting predictions.
Open AccessSelf-supervised Spatial-Temporal Learner for Precipitation Nowcasting
Nowcasting, the short-term prediction of weather, is essential for making timely and weather-dependent decisions. Specifically, precipitation nowcasting aims to predict precipitation at a local level within a 6-hour time frame. This task can be framed as a spatial-temporal sequence forecasting problem, where deep learning methods have been particularly effective. However, despite advancements in self-supervised learning, most successful methods for nowcasting remain fully supervised. Self-supervised learning is advantageous for pretraining models to learn representations without requiring extensive labeled data. In this work, we leverage the benefits of self-supervised learning and integrate it with spatial-temporal learning to develop a novel model, SpaT-SparK. SpaT-SparK comprises a CNN-based encoder-decoder structure pretrained with a masked image modeling (MIM) task and a translation network that captures temporal relationships among past and future precipitation maps in downstream tasks. We conducted experiments on the NL-50 dataset to evaluate the performance of SpaT-SparK. The results demonstrate that SpaT-SparK outperforms existing baseline supervised models, such as SmaAt-UNet, providing more accurate nowcasting predictions.