Mohammed Ouzzif
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
Mohammed Ouzzif has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
- Khalid Bouragba
Laboratory of Computer Science and Smart SystemsESTC, University Hassan II Casablanca, Casablanca, Morocco
5 shared publications - Meriyem Chergui
C3S Laboratory, Higher School of Technology, Hassan II University, Casablanca 20010, Morocco
4 shared publications - Mohamed El Kamili
C3S Laboratory, Higher School of Technology, Hassan II University, Casablanca 20010, Morocco
4 shared publications - Abdellatif Sassioui
C3S Laboratory, Higher School of Technology, Hassan II University, Casablanca 20010, Morocco
3 shared publications - Khalid Bouragba
Hassan II University Casablanca
3 shared publications - Yasser Elouargui
C3S Laboratory, Higher School of Technology, Hassan II University, Casablanca 20010, Morocco
3 shared publications
Latest Publications
- Open Access
- 9 November 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (21) , 11092
TC-HGAT: A Task-Conditioned Hierarchical Graph Attention Network for ESG Text Classification
Classifying corporate environmental, social, and governance (ESG) disclosures is challenging because they combine technical terminology with subtle, domain-specific distinctions. Pretrained Transformers such as ClimateBERT perform well but require many parameters and substantial GPU memory. This paper introduces TC-HGAT, a lightweight hierarchical graph attention network with task-semantic conditioning. TC-HGAT uses frozen FastText-300D embeddings and builds, for each document, a graph of word nodes, virtual cluster nodes, and a task-semantic node obtained by mean-pooling the embeddings of a natural-language task description. On three ESG tasks (Claim, Specificity, and Commitment), TC-HGAT achieves macro-F1 scores of 94.6%, 77.8%, and 79.3%, with the highest mean scores on Claim and Commitment, although no difference from ClimateBERT is statistically significant. Most of the gain comes from the task-semantic node, which improves macro-F1 significantly on all three tasks (by 5.7, 2.4, and 4.5 points), whereas the cluster layer alone gives a modest improvement that is significant only on Claim. Random vectors in the same graph position, including a norm-matched one, do not reproduce this gain, indicating that it depends on the content of the task representation. TC-HGAT trains about 1.4 million parameters, far fewer than BERT-base and ClimateBERT, and needs substantially less peak GPU memory during training.
- 9 November 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (21) , 11092
Open AccessTC-HGAT: A Task-Conditioned Hierarchical Graph Attention Network for ESG Text Classification
Classifying corporate environmental, social, and governance (ESG) disclosures is challenging because they combine technical terminology with subtle, domain-specific distinctions. Pretrained Transformers such as ClimateBERT perform well but require many parameters and substantial GPU memory. This paper introduces TC-HGAT, a lightweight hierarchical graph attention network with task-semantic conditioning. TC-HGAT uses frozen FastText-300D embeddings and builds, for each document, a graph of word nodes, virtual cluster nodes, and a task-semantic node obtained by mean-pooling the embeddings of a natural-language task description. On three ESG tasks (Claim, Specificity, and Commitment), TC-HGAT achieves macro-F1 scores of 94.6%, 77.8%, and 79.3%, with the highest mean scores on Claim and Commitment, although no difference from ClimateBERT is statistically significant. Most of the gain comes from the task-semantic node, which improves macro-F1 significantly on all three tasks (by 5.7, 2.4, and 4.5 points), whereas the cluster layer alone gives a modest improvement that is significant only on Claim. Random vectors in the same graph position, including a norm-matched one, do not reproduce this gain, indicating that it depends on the content of the task representation. TC-HGAT trains about 1.4 million parameters, far fewer than BERT-base and ClimateBERT, and needs substantially less peak GPU memory during training.
- Open Access
- 25 April 2026
- Published by MDPI AG in Technologies
- Vol. 14 (5) , 258
ESG-Graph: Hierarchical Residual Graph Attention Network with Analyst-Defined ESG Taxonomy
Environmental, Social, and Governance (ESG) text classification is important for applications in sustainable finance. However, it remains a challenging task due to domain terminology and regulatory constraints. While transformer-based models achieve strong predictive performance, they often lead to high energy costs and provide limited interpretability. To address these limitations, we introduce ESG-Graph, a lightweight and interpretable graph-based framework for modeling ESG disclosures. In our approach, each sentence is represented as a token-level dependency graph augmented with virtual nodes initialized from a European Sustainability Reporting Standards (ESRS)-based taxonomy, enabling the addition of new ESG concepts without retraining. A multi-layer Graph Attention Network is used instead of transformer encoders, allowing grammatical structure and domain semantics to be modeled jointly. Experiments on three ESG benchmark datasets show that ESG-Graph achieves performance comparable to efficient transformer baselines while consuming up to 60× less energy and using 10× fewer parameters. Additional attribution and ablation studies suggest the method’s policy alignment, interpretability, and robustness.
- 25 April 2026
- Published by MDPI AG in Technologies
- Vol. 14 (5) , 258
Open AccessESG-Graph: Hierarchical Residual Graph Attention Network with Analyst-Defined ESG Taxonomy
Environmental, Social, and Governance (ESG) text classification is important for applications in sustainable finance. However, it remains a challenging task due to domain terminology and regulatory constraints. While transformer-based models achieve strong predictive performance, they often lead to high energy costs and provide limited interpretability. To address these limitations, we introduce ESG-Graph, a lightweight and interpretable graph-based framework for modeling ESG disclosures. In our approach, each sentence is represented as a token-level dependency graph augmented with virtual nodes initialized from a European Sustainability Reporting Standards (ESRS)-based taxonomy, enabling the addition of new ESG concepts without retraining. A multi-layer Graph Attention Network is used instead of transformer encoders, allowing grammatical structure and domain semantics to be modeled jointly. Experiments on three ESG benchmark datasets show that ESG-Graph achieves performance comparable to efficient transformer baselines while consuming up to 60× less energy and using 10× fewer parameters. Additional attribution and ablation studies suggest the method’s policy alignment, interpretability, and robustness.
- Open Access
- 23 April 2026
- Published by MDPI AG in Technologies
- Vol. 14 (5) , 250
AMD-Proj: Adaptive Memory-Driven Selective Gradient Projection for Continual Learning in Document Understanding
Visually rich document understanding (VrDU) models rely on tightly coupled textual, layout, and visual representations. In real-world deployments, these models must continuously adapt to new document domains over time. However, naïve sequential fine-tuning leads to severe catastrophic forgetting due to shared parameters and strong cross-task interference. Existing continual learning approaches either constrain parameter updates, preserve output distributions, or uniformly suppress gradient directions associated with previous tasks. While effective in limited settings, these strategies fail to balance stability and plasticity in large multimodal transformers. We propose AMD-Proj, an adaptive memory-driven selective gradient projection framework for continual learning in document understanding. It models task knowledge using specific gradient subspaces and adaptively modulates incoming gradients based on their alignment with this memory, selectively blocking interfering directions while reinforcing reusable ones. An efficient truncated SVD mechanism with online subspace merging ensures bounded memory usage and scalability to large transformer-based architectures. We evaluate AMD-Proj on four VrDU benchmarks (FUNSD, SROIE, CORD, and BuDDIE) under a task-incremental learning setting using LayoutLMv2 and LayoutLMv3 backbones. Results show that AMD-Proj reduces catastrophic forgetting and improves F1-based stability over EWC, GPM, LwF, OWM, CUBER, TRGP and parameter-efficient fine-tuning methods. Extensive mechanistic analyses, including gradient spectrum decomposition and layer-wise reuse versus block dynamics, provide insight into how selective gradient projection controls optimization geometry during continual adaptation. These findings establish selective gradient projection as a principled and interpretable approach for continual learning in visually rich document understanding.
- 23 April 2026
- Published by MDPI AG in Technologies
- Vol. 14 (5) , 250
Open AccessAMD-Proj: Adaptive Memory-Driven Selective Gradient Projection for Continual Learning in Document Understanding
Visually rich document understanding (VrDU) models rely on tightly coupled textual, layout, and visual representations. In real-world deployments, these models must continuously adapt to new document domains over time. However, naïve sequential fine-tuning leads to severe catastrophic forgetting due to shared parameters and strong cross-task interference. Existing continual learning approaches either constrain parameter updates, preserve output distributions, or uniformly suppress gradient directions associated with previous tasks. While effective in limited settings, these strategies fail to balance stability and plasticity in large multimodal transformers. We propose AMD-Proj, an adaptive memory-driven selective gradient projection framework for continual learning in document understanding. It models task knowledge using specific gradient subspaces and adaptively modulates incoming gradients based on their alignment with this memory, selectively blocking interfering directions while reinforcing reusable ones. An efficient truncated SVD mechanism with online subspace merging ensures bounded memory usage and scalability to large transformer-based architectures. We evaluate AMD-Proj on four VrDU benchmarks (FUNSD, SROIE, CORD, and BuDDIE) under a task-incremental learning setting using LayoutLMv2 and LayoutLMv3 backbones. Results show that AMD-Proj reduces catastrophic forgetting and improves F1-based stability over EWC, GPM, LwF, OWM, CUBER, TRGP and parameter-efficient fine-tuning methods. Extensive mechanistic analyses, including gradient spectrum decomposition and layer-wise reuse versus block dynamics, provide insight into how selective gradient projection controls optimization geometry during continual adaptation. These findings establish selective gradient projection as a principled and interpretable approach for continual learning in visually rich document understanding.
- Open Access
- 1 April 2026
- Published by Institute of Advanced Engineering and Science in Indonesian Journal of Electrical Engineering and Computer Science
- Vol. 42 (1)
A decentralized call recording in voice over IP based on blockchain using smart contracts
Although voice over IP (VoIP) has established itself as the new paradigm for universal telecommunications, its massive deployment within businesses and government agencies has paradoxically increased the attack surface for cyber threats: stream injection fraud, identity theft, and, more recently, the emergence of voice deepfakes, rendering traditional security architectures obsolete. At the same time, conventional centralized recording systems raise trust issues, as they are vulnerable to data manipulation, unauthorized access, and single points of failure. This article presents a new architecture that decentralizes the recording and securing of VoIP calls by combining three key technologies: blockchain for immutability; smart contracts to automate communications governance and ensure the transition from a centralized to an algorithmic trust model; and artificial intelligence (AI) agents that analyze audio streams in real time. This approach transforms VoIP recording from a simple passive file into a secure, auditable, and confidential digital asset. By removing centralized control and strengthening identity verification, this architecture provides a concrete response to security requirements.
- 1 April 2026
- Published by Institute of Advanced Engineering and Science in Indonesian Journal of Electrical Engineering and Computer Science
- Vol. 42 (1)
Open AccessA decentralized call recording in voice over IP based on blockchain using smart contracts
Although voice over IP (VoIP) has established itself as the new paradigm for universal telecommunications, its massive deployment within businesses and government agencies has paradoxically increased the attack surface for cyber threats: stream injection fraud, identity theft, and, more recently, the emergence of voice deepfakes, rendering traditional security architectures obsolete. At the same time, conventional centralized recording systems raise trust issues, as they are vulnerable to data manipulation, unauthorized access, and single points of failure. This article presents a new architecture that decentralizes the recording and securing of VoIP calls by combining three key technologies: blockchain for immutability; smart contracts to automate communications governance and ensure the transition from a centralized to an algorithmic trust model; and artificial intelligence (AI) agents that analyze audio streams in real time. This approach transforms VoIP recording from a simple passive file into a secure, auditable, and confidential digital asset. By removing centralized control and strengthening identity verification, this architecture provides a concrete response to security requirements.
- 26 July 2025
- Published by Springer Nature in Progress in Artificial Intelligence
- p. 1-23
V-FLEX: Verifiable cross-silo federated learning using trusted execution environment
- 26 July 2025
- Published by Springer Nature in Progress in Artificial Intelligence
- p. 1-23
V-FLEX: Verifiable cross-silo federated learning using trusted execution environment
- 14 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 22 (6) , 6592-6609
VeSAFL: Verifiable Secure Aggregation for Privacy-Preserving Federated Learning
With the proliferation of IoT devices and the exponential growth of data generated at the edge, federated learning (FL) emerges as a powerful method for training machine learning models on decentralized data sources. In security-critical applications, such as anomaly and threat detection, ensuring the confidentiality and integrity of sensitive data is paramount. In this paper, we introduce VeSAFL, a novel scheme for verifiable secure aggregation for privacy-preserving FL designed for edge computing environments. VeSAFL decentralizes model updates across edge nodes, reducing dependency on centralized cloud servers and mitigating risks associated with single points of failure. By leveraging multi-server aggregators, our approach fortifies system resilience and reliability against potential cyber threats. To bolster trust in the learning process, we implement a robust verification mechanism that guarantees the integrity and authenticity of local and global updates. Our experimental results highlight the efficacy and efficiency of VeSAFL in safeguarding against active adversaries and accurately identifying anomalous activities. Furthermore, our comprehensive security analysis affirms the scheme's correctness, verifiability, and privacy preservation in adversarial scenarios.
- 14 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 22 (6) , 6592-6609
VeSAFL: Verifiable Secure Aggregation for Privacy-Preserving Federated Learning
With the proliferation of IoT devices and the exponential growth of data generated at the edge, federated learning (FL) emerges as a powerful method for training machine learning models on decentralized data sources. In security-critical applications, such as anomaly and threat detection, ensuring the confidentiality and integrity of sensitive data is paramount. In this paper, we introduce VeSAFL, a novel scheme for verifiable secure aggregation for privacy-preserving FL designed for edge computing environments. VeSAFL decentralizes model updates across edge nodes, reducing dependency on centralized cloud servers and mitigating risks associated with single points of failure. By leveraging multi-server aggregators, our approach fortifies system resilience and reliability against potential cyber threats. To bolster trust in the learning process, we implement a robust verification mechanism that guarantees the integrity and authenticity of local and global updates. Our experimental results highlight the efficacy and efficiency of VeSAFL in safeguarding against active adversaries and accurately identifying anomalous activities. Furthermore, our comprehensive security analysis affirms the scheme's correctness, verifiability, and privacy preservation in adversarial scenarios.
- Open Access
- 9 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 84569-84579
LMGDoc: Light Multigranular GNN for Efficient Document Understanding
Document understanding is a critical task in extracting structured information from documents such as forms, receipts, and reports. Visually Rich Documents (VRDs) present unique challenges due to their complex layouts, heterogeneous content (text, tables, images), and diverse structural patterns. While recent transformer-based models have achieved strong performance, they typically rely on large-scale pretraining and have high computational demands, limiting their usability in real-time or resourceconstrained settings. In this paper, we propose LMGDoc, a lightweight model capable of understanding and processing VRDs composed of three main components: a Feature Extractor using GloVe for text embedding and a novel layout encoding method, a Graph Constructor for representing document structures at multiple granular levels (word, region, and page), and a Graph Convolution Network (GNN). LMGDoc achieves competitive results on classification and Key Information Extraction (KIE) benchmarks with only 0.4M parameters size compared to state-of-the-art models.
- 9 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 84569-84579
Open AccessLMGDoc: Light Multigranular GNN for Efficient Document Understanding
Document understanding is a critical task in extracting structured information from documents such as forms, receipts, and reports. Visually Rich Documents (VRDs) present unique challenges due to their complex layouts, heterogeneous content (text, tables, images), and diverse structural patterns. While recent transformer-based models have achieved strong performance, they typically rely on large-scale pretraining and have high computational demands, limiting their usability in real-time or resourceconstrained settings. In this paper, we propose LMGDoc, a lightweight model capable of understanding and processing VRDs composed of three main components: a Feature Extractor using GloVe for text embedding and a novel layout encoding method, a Graph Constructor for representing document structures at multiple granular levels (word, region, and page), and a Graph Convolution Network (GNN). LMGDoc achieves competitive results on classification and Key Information Extraction (KIE) benchmarks with only 0.4M parameters size compared to state-of-the-art models.
- Open Access
- 3 September 2024
- Published by MDPI AG in Computation
- Vol. 12 (9) , 179
PyIncentiveBC: A Python Module for Simulation of Incentivization Mechanism Implemented in Blockchain-Based Systems
The diversity of approaches for retaining participants in a Blockchain-based system complicates benchmarking. The majority of proposals for rewarding and penalizing participants in these systems are limited to their own set of data and scenarios, making it hard to compare their effectiveness. To overcome these challenges, we developed PyIncentiveBC, a free, open-source, and modular simulator designed to evaluate the reliability of any approach, incorporating a dynamic and proportionate incentivization mechanism proposed in our previous work. PyIncentiveBC aims to provide the scientific communities with an extensible software solution facilitating the benchmarking of existing approaches with new ones proposed by them.
- 3 September 2024
- Published by MDPI AG in Computation
- Vol. 12 (9) , 179
Open AccessPyIncentiveBC: A Python Module for Simulation of Incentivization Mechanism Implemented in Blockchain-Based Systems
The diversity of approaches for retaining participants in a Blockchain-based system complicates benchmarking. The majority of proposals for rewarding and penalizing participants in these systems are limited to their own set of data and scenarios, making it hard to compare their effectiveness. To overcome these challenges, we developed PyIncentiveBC, a free, open-source, and modular simulator designed to evaluate the reliability of any approach, incorporating a dynamic and proportionate incentivization mechanism proposed in our previous work. PyIncentiveBC aims to provide the scientific communities with an extensible software solution facilitating the benchmarking of existing approaches with new ones proposed by them.
- 4 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Engineering Management Review
- Vol. 52 (2) , 55-68
A Smart Contract Architecture Framework for Insurance Industry Using Blockchain and Business Process Management Technology
The Covid-19 pandemic prompted insurance companies to shift from their traditional operating mode to a digital one. This shift is to promote improved performance, cost reduction, and enhanced customer service. Blockchain is to aid the insurance industry radically transform its operations by enabling faster exchange of verifiable data, visibility for all parties, and transactions underpinned by pervasive security and trust. It permits the replacement of third parties so as to benefit from timestamping services, storage of evidence and the history of actions taken to verify the identity of the original issuer and any co-signers of a document. In this paper we propose a solution for insurance service providers based on a private blockchain for Einsurance service providers-customers corresponding to a community network. The objective of this paper is to support the actors involved in this decision-making process by illustrating what a blockchain is, by analyzing its advantages and disadvantages, as well as by discussing the use cases drawn from the sector of insurance, which could easily be extended to another domain. Besides, we present our architecture proposal to make all types of insurance smart contracts based on BPM (Business Process Management) and Blockchain.
- 4 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Engineering Management Review
- Vol. 52 (2) , 55-68
A Smart Contract Architecture Framework for Insurance Industry Using Blockchain and Business Process Management Technology
The Covid-19 pandemic prompted insurance companies to shift from their traditional operating mode to a digital one. This shift is to promote improved performance, cost reduction, and enhanced customer service. Blockchain is to aid the insurance industry radically transform its operations by enabling faster exchange of verifiable data, visibility for all parties, and transactions underpinned by pervasive security and trust. It permits the replacement of third parties so as to benefit from timestamping services, storage of evidence and the history of actions taken to verify the identity of the original issuer and any co-signers of a document. In this paper we propose a solution for insurance service providers based on a private blockchain for Einsurance service providers-customers corresponding to a community network. The objective of this paper is to support the actors involved in this decision-making process by illustrating what a blockchain is, by analyzing its advantages and disadvantages, as well as by discussing the use cases drawn from the sector of insurance, which could easily be extended to another domain. Besides, we present our architecture proposal to make all types of insurance smart contracts based on BPM (Business Process Management) and Blockchain.
- Open Access
Vulnerability Detection Approaches on Application Behaviors in Mobile Environment
Several solutions ensuring the dynamic detection of malicious activities on Android ecosystem have been proposed. These are represented by generic rules and models that identify any purported malicious behavior. However, the approaches adopted are far from being effective in detecting malware (listed or not) and whose form and behavior are likely to be different depending on the execution environment or the design of the malware itself (polymorphic for example). An additional difficulty is added when these approaches are unable to capture, analyze, and classify all the execution paths incorporated in the analyzed application earlier. This suggests that the functionality of the analyzed application can constitute a potential risk but never explored or revealed. We have studied some malware detection techniques based on behavioral analysis of applications. The description, characteristics, and results obtained from each technique are presented in this article wherein we have also highlighted some open problems, challenges as well as the different possible future directions of research concerning behavioral analysis of malware.
Open AccessVulnerability Detection Approaches on Application Behaviors in Mobile Environment
Several solutions ensuring the dynamic detection of malicious activities on Android ecosystem have been proposed. These are represented by generic rules and models that identify any purported malicious behavior. However, the approaches adopted are far from being effective in detecting malware (listed or not) and whose form and behavior are likely to be different depending on the execution environment or the design of the malware itself (polymorphic for example). An additional difficulty is added when these approaches are unable to capture, analyze, and classify all the execution paths incorporated in the analyzed application earlier. This suggests that the functionality of the analyzed application can constitute a potential risk but never explored or revealed. We have studied some malware detection techniques based on behavioral analysis of applications. The description, characteristics, and results obtained from each technique are presented in this article wherein we have also highlighted some open problems, challenges as well as the different possible future directions of research concerning behavioral analysis of malware.