AFAicha Beya Far

Aicha Beya Far

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

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


Aicha Beya Far has a total of 31 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Artificial Intelligence Across the Radiology Workflow: A Nine-Stage Narrative Review

    by Marwa Chendeb El Rai, Aicha Beya Far, Muna Darweesh, Salam Dhou, Nour Aburaed, Salah El Rai, Mohammed ElKhazendar, Samer Ellahham

    Radiology services are experiencing increasing operational complexity due to rising imaging volumes and expanding coordination demands across interconnected clinical and administrative processes. This complexity is reflected in variability across workflow stages, driven by fragmented information flows, heterogeneous system integration, and multi-source data dependencies. Artificial intelligence (AI) has therefore emerged as a potential tool to support automation, prioritization, and operational efficiency throughout the radiology pathway. This narrative review examines published applications of AI within a nine-stage representation of the radiology workflow. The review synthesizes how AI methods are being investigated to support both administrative coordination and diagnostic processes in radiology practice. AI approaches aim to reduce repetitive administrative tasks, improve resource utilization, and assist radiologists in managing increasing imaging workloads. However, research activity remains uneven, with a strong concentration on later-stage tasks such as image analysis and reporting, while earlier and administrative stages remain comparatively underexplored. By organizing existing research within a unified workflow-oriented framework, this review highlights areas of concentration and identifies gaps across less-studied stages. The findings suggest that while several AI applications are approaching early clinical deployment, broader workflow-level impact remains limited by challenges related to system integration, interoperability, governance, and real world implementation. Continued progress will depend on developing integrated and clinically validated solutions that extend beyond isolated tasks to support coordinated radiology workflow optimization.


  • Open Access

    AT-HSTNet: An Efficient Hierarchical Action-Transformer Framework for Deepfake Video Detection

    by Sameena Javaid, Marwa Chendeb El Rai, Abeer Elkhouly, Obada Al-Khatib, Aicha Beya Far, May El Barachi

    The rapid advancement of deepfake generation technologies presents significant challenges to the verification of digital video authenticity. These time-dependent artifacts are difficult to detect using conventional frame-based detection approaches. This paper introduces AT-HSTNet, an Action-Transformer-based Hierarchical Spatiotemporal Network designed for robust and computationally efficient deepfake video detection. The proposed framework adopts a multi-stage hierarchical architecture in which frame-level visual features are extracted using an EfficientNet-B0 backbone, short- and medium-range temporal patterns are modeled through Bidirectional Long Short-Term Memory (BiLSTM) networks, and long-range temporal dependencies are captured using an action-aware Transformer operating on temporally aggregated representations. Unlike conventional video transformers that apply self-attention directly to raw frame-level features, the proposed action-aware attention mechanism reduces redundant computation and improves stability in temporal reasoning. Extensive experiments on the balanced FFIW-10K dataset demonstrate that AT-HSTNet achieves an accuracy of 98.7%, with 98.0% precision, 96.0% recall, and a 96.9% F1-score, outperforming representative CNN–BiLSTM and CNN–Transformer baseline architectures. In addition, AT-HSTNet is highly efficient, requiring only 0.45 GFLOPs and achieving an inference speed of approximately 30 FPS on consumer-grade GPU hardware. As a result of this study, we found hierarchical temporal modeling more effective when combined with action-aware attention for any deepfake video detection.


  • Open Access

    Machine Learning for Wearable Sensor-Based Human Movement Rehabilitation: A Five-Year Systematic Review

    by Yassine Benachour, Farid Flitti, Lina Maloukh, Aicha Beya Far, Elhocine Boutellaa, Mohamed Bentoumi, Marwa Chendeb El Rai, Nour Aburaed, Khaled Ali, Moez Rehman, et al.


  • Open Access

    Spectral and Spatial Attention Fusion for Building Segmentation in Remote Sensing Imagery

    by Marwa Chendeb El Rai, Muna Darweesh, Aicha Beya Far, Amjad Gawanmeh

    The building segmentation in very high resolution remote sensing imagery presents challenges due to the need to delineate features accurately in a wide range of urban landscapes. Many existing building segmentation methods struggle in discerning complex structures and providing fine grained generalisation over different geographic regions. Additionally, these methods often require to extensive preprocessing and struggle to combine multispectral data. Addressing the different challenges, we introduce the Multi-Band Spectral-Spatial Fusion Attention Network (MBSSFA-Net), a novel method for semantic segmentation. MBSSFA-Net implements a dual encoder designed to exploit the complementary spectral information provided by the Near-Infrared and the RGB bands, to improve the feature representation and the segmentation accuracy. The approach incorporates multiscale spectral and spatial attention fusion blocks in the encoder to fuse the extracted features to enhance boundary delineation, and a spectral and spatial attention fusion blocks in the decoder to merge the spatial and abstract features. The proposed framework can extract buildings in different environment since it can process multispectral data. The experiments have been performed on GaoFen-7 and WHU-Satellite II datasets. The experiments prove that our method outperforms current state of the art Deep Learning segmentation techniques, demonstrating its potential for building segmentation in complex urban environments.


  • A Bio-Inspired Pattern Recognition System for Tin-Oxide Gas Sensor Applications

    by Aicha Beya Far, Farid Flitti, Bin Guo, Amine Bermak

    In this paper, a bio-inspired pattern recognition system for tin-oxide gas sensor applications is proposed. To mimic the biological olfactory system, temperature modulation is first used to virtually increase the number of sensors by periodically sampling the sensors' response at different temperatures. A convex microhotplate is used in order to improve the thermal properties of the structure enabling efficient temperature modulation process to be carried out. Temperature modulation is shown to increase the number of effective sensors from 16 physically available sensors (integrated on a single chip) to 12 000 virtual sensors (VSs). This enables the emulation of a very large number of sensors typically found in biological systems. The response of each sensor is seen as a fingerprint map, which is further processed using various image processing techniques. Self organized maps (SOMs) algorithm is used to create a 2D map for each gas and to combine the huge number of VSs in order to reduce the dimensionality. Image moments are used as a feature enabling to characterize the spatial distribution within the image lattice and to retrieve the brighter regions in the SOM' nodes exhibiting high activity to the input gas. Experiments on real sensors data show improved performance (96%) as compared with standard gas discrimination algorithms.


  • Gas identification system based on temperature modulation tin-oxide sensors and bio-inspired processing

    by A. Far, F. Flitti, B. Guo, A. Bermak

    This paper presents a gas identification system for electronic nose inspired from the biological olfactory system. Temperature modulation is first used to virtually boost the number of sensors by periodically sampling the sensorspsila response at different temperatures. Convex tin-oxide sensor technology is used to increase the thermal efficiency of our in-house gas sensors. The response of each sensor is seen as a fingerprint map which is further processed using various image processing techniques. Indeed, SOM algorithm is used to create a 2D map for each gas and the image moments are calculated to retrieve the blighter regions in the SOMpsilas nodes that show high activity to the input gas. The obtained results using the proposed technique are very encouraging.


  • Open Access

    A Robust and Low‐Complexity Gas Recognition Technique for On‐Chip Tin‐Oxide Gas Sensor Array

    by Farid Flitti, Aicha Far, Bin Guo, Amine Bermak

    Gas recognition is a new emerging research area with many civil, military, and industrial applications. The success of any gas recognition system depends on its computational complexity and its robustness. In this work, we propose a new low‐complexity recognition method which is tested and successfully validated for tin‐oxide gas sensor array chip. The recognition system is based on a vector angle similarity measure between the query gas and the representatives of the different gas classes. The latter are obtained using a clustering algorithm based on the same measure within the training data set. Experimented results on our in‐house gas sensors array show more than 98% of correct recognition. The robustness of the proposed method is tested by recognizing gas measurements with simulated drift. Less than 1% of performance degradation is noted at the worst case scenario which represents a significant improvement when compared to the current state‐of‐the‐art.


  • Drift Invariant Gas Recognition Technique for On Chip Tin Oxide Gas Sensor Array

    by F. Flitti, A. Far, B. Guo, A. Bermak

    The purpose of this paper is the study of the robustness of a new low complexity recognition method based on the measurement issued from an on chip 4times4 Tin oxide gas sensor array. The recognition system is based on a vector angle similarity measure between the query gas and the representatives of the different gas classes. The latter are obtained using a clustering algorithm based on the same measure within the training data set. Experimented results show more than 98% of good recognition and the robustness of the proposed approach is tested by recognizing gas measurements with simulated drift. Less than 1% of performance degradation is noted at the worst case.


  • Temperature Modulation for Tin-Oxide Gas Sensors

    by A. Far, B. Guo, F. Flitti, A. Bermak

    This paper presents a study of temperature modulation for tin-oxide gas sensor. The main objective of this work is to perform temperature modulation experimental setup for tin oxide gas sensors in order to improve the selectivity of the sensor array and to virtually increase the number of sensors. Typically, tin oxide sensors operate by heating at a relatively high temperature (around 300degC a microhotplate structure). A convex microhotplate is proposed in order to improve the thermal properties of the structure and enable efficient temperature modulation process to be carried-out. Temperature modulation is shown to increase the number of our sensors from 16 physical sensors (integrated on-chip) up to 12 000 virtual sensors. This will enable the emulation of a very large number of sensors typically found in biological systems.


  • A new Gas Recognition Technique using On Chip Sensor Array Measurements

    by F. Flitti, B. Guo, A. Far, A. Bermak

    This paper deals with gas recognition which has many applications in various domestic and industry tasks. We propose a new low complexity recognition method based on the measurement issued from an on chip 4 x 4 gas sensor array. The recognition system is based on a vector angle similarity measure between the query gas and the representatives of the different gas classes. The latter are obtained using a clustering algorithm based on the same measure within the training data set. Experimented results show more than 98% of good recognition.