Sareh Soleimani
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Sareh Soleimani has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- Hussein Al Osman
School of Electrical Engineering and Computer Science, University of Ottawa, Ottawa, ON, Canada
2 shared publications - Frank Knoefel
Memory Care Research Group, Bruyère Health Research Institute, Ottawa, ON, Canada
1 shared publication - Martin Bouchard
School of Electrical Engineering and Computer Science, University of Ottawa, Ottawa, ON, Canada
1 shared publication - Mohammadreza Sadeghi
Department of Electrical and Computer Engineering, McGill University, Montreal, QC, Canada
1 shared publication - Narges Armanfard
McGill University
1 shared publication - Rafik Goubran
Department of Systems and Computer Engineering, Carleton University, Ottawa, ON, Canada
1 shared publication
Latest Publications
- 29 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-19
Transformer-Based Bimodal Emotion Recognition: Signal Measurement and Transformer Fusion Approach
Human emotion plays a fundamental role in cognition, decision-making, and social interactions. Electroencephalography (EEG) is a fundamental widely used neurophysiological measurement tool that records brain electrical activity, providing valuable insights into emotional states. This paper presents a novel transformer-based bimodal emotion recognition model that integrates EEG and facial expression data to enhance multimodal representation learning and emotion classification accuracy. A comprehensive evaluation is conducted by implementing and benchmarking various state-of-the-art unimodal and bimodal emotion recognition models as well as exploring multiple fusion techniques. The model’s performance is validated on two widely used datasets, DEAP and MAHNOB-HCI, where the proposed model outperforms existing approaches in both unimodal and bimodal configurations, achieving an approximate 4% higher accuracy for valence and arousal dimensions. Specifically, for the DEAP dataset, the model attains classification accuracies of 0.66 (valence) and 0.72 (arousal) on the training set and 0.65 (valence) and 0.66 (arousal) on the test set. For the MAHNOB-HCI dataset, it achieves 0.69 (valence) and 0.73 (arousal) on the training set and 0.61 (valence) and 0.68 (arousal) on the test set. Additionally, the proposed transformer-based fusion method outperforms traditional feature-level and decision-level fusion techniques.
- 29 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-19
Transformer-Based Bimodal Emotion Recognition: Signal Measurement and Transformer Fusion Approach
Human emotion plays a fundamental role in cognition, decision-making, and social interactions. Electroencephalography (EEG) is a fundamental widely used neurophysiological measurement tool that records brain electrical activity, providing valuable insights into emotional states. This paper presents a novel transformer-based bimodal emotion recognition model that integrates EEG and facial expression data to enhance multimodal representation learning and emotion classification accuracy. A comprehensive evaluation is conducted by implementing and benchmarking various state-of-the-art unimodal and bimodal emotion recognition models as well as exploring multiple fusion techniques. The model’s performance is validated on two widely used datasets, DEAP and MAHNOB-HCI, where the proposed model outperforms existing approaches in both unimodal and bimodal configurations, achieving an approximate 4% higher accuracy for valence and arousal dimensions. Specifically, for the DEAP dataset, the model attains classification accuracies of 0.66 (valence) and 0.72 (arousal) on the training set and 0.65 (valence) and 0.66 (arousal) on the test set. For the MAHNOB-HCI dataset, it achieves 0.69 (valence) and 0.73 (arousal) on the training set and 0.61 (valence) and 0.68 (arousal) on the test set. Additionally, the proposed transformer-based fusion method outperforms traditional feature-level and decision-level fusion techniques.
- Open Access
- 20 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 148104-148127
Deep Clustering With Self-Supervision Using Pairwise Similarities
Deep clustering incorporates embedding into clustering to find a lower-dimensional space appropriate for clustering. In this paper, we propose a novel deep clustering framework with self-supervision using pairwise data similarities (DCSS). The proposed method consists of two successive phases. In the first phase, we propose to form hypersphere-like groups of similar data points, i.e. one hypersphere per cluster, employing an autoencoder that is trained using cluster-specific losses. The hyper-spheres are formed in the autoencoder’s latent space. In the second phase, we propose to employ pairwise data similarities to create a K-dimensional space that is capable of accommodating more complex cluster distributions, hence providing more accurate clustering performance. K is the number of clusters. The autoencoder’s latent space obtained in the first phase is used as the input of the second phase. The effectiveness of both phases is demonstrated on seven benchmark datasets by conducting a rigorous set of experiments. The DCSS code is available: https://github.com/Armanfard-Lab/DCSS.
- 20 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 148104-148127
Open AccessDeep Clustering With Self-Supervision Using Pairwise Similarities
Deep clustering incorporates embedding into clustering to find a lower-dimensional space appropriate for clustering. In this paper, we propose a novel deep clustering framework with self-supervision using pairwise data similarities (DCSS). The proposed method consists of two successive phases. In the first phase, we propose to form hypersphere-like groups of similar data points, i.e. one hypersphere per cluster, employing an autoencoder that is trained using cluster-specific losses. The hyper-spheres are formed in the autoencoder’s latent space. In the second phase, we propose to employ pairwise data similarities to create a K-dimensional space that is capable of accommodating more complex cluster distributions, hence providing more accurate clustering performance. K is the number of clusters. The autoencoder’s latent space obtained in the first phase is used as the input of the second phase. The effectiveness of both phases is demonstrated on seven benchmark datasets by conducting a rigorous set of experiments. The DCSS code is available: https://github.com/Armanfard-Lab/DCSS.
- Open Access
- 18 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 113-130
A Graph Neural Network for EEG-Based Emotion Recognition With Contrastive Learning and Generative Adversarial Neural Network Data Augmentation
The limited size of existing datasets and signal variability have hindered EEG-based emotion recognition. In this paper, we present a solution that simultaneously addresses both problems. Generative Adversarial Networks (GANs) have recently shown notable data augmentation (DA) success. Therefore, we leverage a GAN-based DA technique to enhance the robustness of our proposed emotion recognition model by synthetically increasing the size of our datasets. Moreover, we employ contrastive learning to improve the quality of the learned representations from EEG signals and mitigate the adverse impact of inter-subject and intra-subject variability in signals corresponding to the same stimuli or emotions. We do so by maximizing the similarity in the representation of such EEG signals. We perform EEG-based emotion classification using a Graph Neural Network (GNN), which learns the relationship between the extracted EEG features. We compare the proposed model with several recent state-of-the-art emotion recognition models on the DEAP and MAHNOB datasets. The experimental results demonstrate that the proposed model outperforms previous models with a 64.84% and 66.40% emotion classification accuracy on the test set of the DEAP dataset and a 66.98% and 71.69% emotion classification accuracy on the test set of the MAHNOB-HCI dataset for the valence and arousal emotional dimensions, respectively. We perform an ablation study to demonstrate how contrastive learning, GAN, and GNN contribute to improving the proposed solution’s performance.
- 18 December 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 113-130
Open AccessA Graph Neural Network for EEG-Based Emotion Recognition With Contrastive Learning and Generative Adversarial Neural Network Data Augmentation
The limited size of existing datasets and signal variability have hindered EEG-based emotion recognition. In this paper, we present a solution that simultaneously addresses both problems. Generative Adversarial Networks (GANs) have recently shown notable data augmentation (DA) success. Therefore, we leverage a GAN-based DA technique to enhance the robustness of our proposed emotion recognition model by synthetically increasing the size of our datasets. Moreover, we employ contrastive learning to improve the quality of the learned representations from EEG signals and mitigate the adverse impact of inter-subject and intra-subject variability in signals corresponding to the same stimuli or emotions. We do so by maximizing the similarity in the representation of such EEG signals. We perform EEG-based emotion classification using a Graph Neural Network (GNN), which learns the relationship between the extracted EEG features. We compare the proposed model with several recent state-of-the-art emotion recognition models on the DEAP and MAHNOB datasets. The experimental results demonstrate that the proposed model outperforms previous models with a 64.84% and 66.40% emotion classification accuracy on the test set of the DEAP dataset and a 66.98% and 71.69% emotion classification accuracy on the test set of the MAHNOB-HCI dataset for the valence and arousal emotional dimensions, respectively. We perform an ablation study to demonstrate how contrastive learning, GAN, and GNN contribute to improving the proposed solution’s performance.
- Open Access
- 11 October 2018
- Published by Association for Computing Machinery (ACM)
- p. 23-35
Long-Term Sleep Assessment by Unobtrusive Pressure Sensor Arrays
- 11 October 2018
- Published by Association for Computing Machinery (ACM)
- p. 23-35
Open AccessLong-Term Sleep Assessment by Unobtrusive Pressure Sensor Arrays