HOHamza Ouarnoughi

Hamza Ouarnoughi

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Hamza Ouarnoughi has a total of 35 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    SONATA: Self-adaptive Evolutionary Framework for Hardware-aware Neural Architecture Search

    by Halima Bouzidi, Smail Niar, Hamza Ouarnoughi, El-Ghazali Talbi

  • Open Access

    Hybrid Predictive Models for Demand and Commodity Price Forecasting in Supply Chains: A Machine Learning Perspective

    by Thi Thuy Hanh Nguyen, Soufiane Douimia, Abdelghani Bekrar, Hamza Ouarnoughi

    Despite extensive development of individual hybrid forecasting models, no systematic comparison of hybridization strategies exists across standardized commodity datasets. This study evaluates twelve hybrid model configurations spanning three frameworks- Combination, Weighting, and Decomposition on two agricultural commodity time series: Vietnamese coffee demand (monthly,2000- 2020) and global wheat prices (daily, 1975 -2025). Results demonstrate that weight-based hybridization consistently outperforms alternatives, delivering 6.9% RMSE improvement over the best standalone baseline in both domains. An ETS-LSTM weighted model achieves MAPE of 9.14% for coffee demand; an ARIMAX-LSTM weighted model attains 8.30% for wheat price. Counterintuitively, the Combination approach theoretically grounded by Zhang[1]- consistently underperforms standalone models, with degradations up to 35.6% for coffee, contradicting the linear-plus-nonlinear decomposition assumption for agricultural commodity series. Structured interviews with nine Vietnamese SMEs identify financial constraints and technical expertise as primary adoption barriers. These findings establish empirically grounded selection guidelines for hybrid forecasting strategies in supply chain contexts.

  • Placement autonomique de machines virtuelles sur un système de stockage hybride dans un cloud IaaS

    by Hamza Ouarnoughi

    Les opérateurs de cloud IaaS (Infrastructure as a Service) proposent à leurs clients des ressources virtualisées (CPU, stockage et réseau) sous forme de machines virtuelles (VM). L’explosion du marché du cloud les a contraints à optimiser très finement l’utilisation de leurs centres de données afin de proposer des services attractifs à moindre coût. En plus des investissements liés à l’achat des infrastructures et de leur coût d’utilisation, la consommation énergétique apparaît comme un point de dépense important (2% de la consommation mondiale) et en constante augmentation. Sa maîtrise représente pour ces opérateurs un levier très intéressant à exploiter. D’un point de vue technique, le contrôle de la consommation énergétique s’appuie essentiellement sur les méthodes de consolidation. Or la plupart d'entre elles ne prennent en compte que l’utilisation CPU des machines physiques (PM) pour le placement de VM. En effet, des études récentes ont montré que les systèmes de stockage et les E/S disque constituent une part considérable de la consommation énergétique d’un centre de données (entre 14% et 40%). Dans cette thèse nous introduisons un nouveau modèle autonomique d’optimisation de placement de VM inspiré de MAPE-K (Monitor, Analyze, Plan, Execute, Knowledge), et prenant en compte en plus du CPU, les E/S des VM ainsi que les systèmes de stockage associés. Ainsi, notre première contribution est relative au développement d’un outil de trace des E/S de VM multi-niveaux. Les traces collectées alimentent, dans l’étape Analyze, un modèle de coût étendu dont l’originalité consiste à prendre en compte le profil d’accès des VM, les caractéristiques du système de stockage, ainsi que les contraintes économiques de l’environnement cloud. Nous analysons par ailleurs les caractéristiques des deux principales classes de stockage, pour aboutir à un modèle hybride exploitant au mieux les avantages de chacune. En effet, les disques durs magnétiques (HDD) sont des supports de stockage à la fois énergivores et peu performants comparés aux unités de calcul. Néanmoins, leur prix par gigaoctet et leur longévité peuvent jouer en leur faveur. Contrairement aux HDD, les disques SSD à base de mémoire flash sont plus performants et consomment peu d’énergie. Leur prix élevé par gigaoctet et leur courte durée de vie (comparés aux HDD) représentent leurs contraintes majeures. L’étape Plan a donné lieu, d’une part, à une extension de l'outil de simulation CloudSim pour la prise en compte des E/S des VM, du caractère hybride du système de stockage, ainsi que la mise en oeuvre du modèle de coût proposé dans l'étape Analyze. Nous avons proposé d’autre part, plusieurs heuristiques se basant sur notre modèle de coût et que nous avons intégrées dans CloudSim. Nous montrons finalement que notre approche permet d’améliorer d’un facteur trois le coût de placement de VM obtenu par les approches existantes.

  • Open Access

    Deep Learning for Anomaly Detection in Railway Systems: A Structured Survey

    by Ammar Bouketta, Smail Niar, Hamza Ouarnoughi

    Ensuring safe and reliable operation of modern railway systems increasingly relies on data-driven monitoring and intelligent fault detection solutions. In recent years, deep learning has emerged as an effective paradigm for anomaly detection in railway applications, driven by the growing availability of heterogeneous sensor data collected from rolling stock and infrastructure. This paper presents a comprehensive survey of deep learning–based anomaly detection approaches developed for railway systems. The surveyed methods are analyzed and organized using a unified taxonomy that captures the location of anomaly occurrence, data representation and manifestation, sensing modality, and temporal characteristics. Existing approaches are further structured into major learning paradigms, including classification-based, prediction-based, reconstruction-based, and hybrid strategies, and their respective strengths and limitations are examined with respect to data availability, real-time constraints, and deployment requirements. Beyond algorithmic design, this survey discusses key data-centric challenges, such as preprocessing, class imbalance, and dataset heterogeneity, and reviews adopted evaluation practices and performance metrics. Practical deployment aspects are also addressed, including edge–cloud architectures, computational constraints, and hardware-aware optimization strategies for embedded railway applications. Building on this analysis, the survey introduces a decision-oriented framework that links anomaly characteristics, data properties, and operational constraints to appropriate detection paradigms and deployment configurations. By jointly considering learning models, data characteristics, evaluation protocols, and deployment constraints, this work provides engineers and researchers with a structured and application-oriented reference for selecting and deploying deep learning solutions for railway anomaly detection, while highlighting open challenges and future research directions toward reliable and scalable intelligent monitoring systems.

  • Open Access

    Combining Neural Architecture Search and Automatic Code Optimization: A Survey

    by Inas Bachiri, Hadjer Benmeziane, Smail Niar, Riyadh Baghdadi, Hamza Ouarnoughi, Abdelkrime Aries

    Deep Learning models have experienced exponential growth in complexity and resource demands in recent years. Accelerating these models for efficient execution on resource-constrained devices has become more crucial than ever. Two notable techniques employed to achieve this goal are Hardware-aware Neural Architecture Search (HW-NAS) and Automatic Code Optimization (ACO). HW-NAS automatically designs accurate yet hardware-friendly neural networks, while ACO involves searching for the best compiler optimizations to apply on neural networks for efficient mapping and inference on the target hardware. This survey explores recent works that combine these two techniques within a single framework. We present the fundamental principles of both domains and demonstrate their sub-optimality when performed independently. We then investigate their integration into a joint optimization process that we call Hardware Aware-Neural Architecture and Compiler Optimizations co-Search (NACOS).

  • Open Access

    FLASH-RL: Federated Learning Addressing System and Static Heterogeneity using Reinforcement Learning

    by Sofiane Bouaziz, Hadjer Benmeziane, Youcef Imine, Leila Hamdad, Smail Niar, Hamza Ouarnoughi

    We propose FLASH-RL, a framework utilizing Double Deep Q-Learning (DDQL) to address system and static heterogeneity in Federated Learning (FL). FLASH-RL introduces a new reputation-based utility function to evaluate client contributions based on their current and past performances. Additionally, an adapted DDQL algorithm is proposed to expedite the learning process. Experimental results on MNIST and CIFAR-10 datasets demonstrate that FLASH-RL strikes a balance between model performance and end-to-end latency, reducing latency by up to 24.83% compared to FedAVG and 24.67% compared to FAVOR. It also reduces training rounds by up to 60.44% compared to FedAVG and 76% compared to FAVOR. Similar improvements are observed on the MobiAct Dataset for fall detection, underscoring the real-world applicability of our approach.


  • Open Access

    Harmonic-NAS: Hardware-Aware Multimodal Neural Architecture Search on Resource-constrained Devices

    by Mohamed Imed Eddine Ghebriout, Halima Bouzidi, Smail Niar, Hamza Ouarnoughi

    The recent surge of interest surrounding Multimodal Neural Networks (MM-NN) is attributed to their ability to effectively process and integrate multiscale information from diverse data sources. MM-NNs extract and fuse features from multiple modalities using adequate unimodal backbones and specific fusion networks. Although this helps strengthen the multimodal information representation, designing such networks is labor-intensive. It requires tuning the architectural parameters of the unimodal backbones, choosing the fusing point, and selecting the operations for fusion. Furthermore, multimodality AI is emerging as a cutting-edge option in Internet of Things (IoT) systems where inference latency and energy consumption are critical metrics in addition to accuracy. In this paper, we propose Harmonic-NAS, a framework for the joint optimization of unimodal backbones and multimodal fusion networks with hardware awareness on resource-constrained devices. Harmonic-NAS involves a two-tier optimization approach for the unimodal backbone architectures and fusion strategy and operators. By incorporating the hardware dimension into the optimization, evaluation results on various devices and multimodal datasets have demonstrated the superiority of Harmonic-NAS over state-of-the-art approaches achieving up to 10.9% accuracy improvement, 1.91x latency reduction, and 2.14x energy efficiency gain.

  • Open Access

    HyT-NAS: Hybrid Transformers Neural Architecture Search for Edge Devices

    by Lotfi Abdelkrim Mecharbat, Hadjer Benmeziane, Hamza Ouarnoughi, Smail Niar

    Vision Transformers have enabled recent attention-based Deep Learning (DL) architectures to achieve remarkable results in Computer Vision (CV) tasks. However, due to the extensive computational resources required, these architectures are rarely implemented on resource-constrained platforms. Current research investigates hybrid handcrafted convolution-based and attention-based models for CV tasks such as image classification and object detection. In this paper, we propose HyT-NAS, an efficient Hardware-aware Neural Architecture Search (HW-NAS) including hybrid architectures targeting vision tasks on tiny devices. HyT-NAS improves state-of-the-art HW-NAS by enriching the search space and enhancing the search strategy as well as the performance predictors. Our experiments show that HyT-NAS achieves a similar hypervolume with less than 5x training evaluations. Our resulting architecture outperforms MLPerf MobileNetVl by 6.3% accuracy improvement with 3.5x less number of parameters on Visual Wake Words.


  • Open Access

    MaGNAS: A Mapping-Aware Graph Neural Architecture Search Framework for Heterogeneous MPSoC Deployment

    by Mohanad Odema, Halima Bouzidi, Hamza Ouarnoughi, Smail Niar, Mohammad Abdullah Al Faruque

    Graph Neural Networks (GNNs) are becoming increasingly popular for vision-based applications due to their intrinsic capacity in modeling structural and contextual relations between various parts of an image frame. On another front, the rising popularity of deep vision-based applications at the edge has been facilitated by the recent advancements in heterogeneous multi-processor Systems on Chips (MPSoCs) that enable inference under real-time, stringent execution requirements. By extension, GNNs employed for vision-based applications must adhere to the same execution requirements. Yet contrary to typical deep neural networks, the irregular flow of graph learning operations poses a challenge to running GNNs on such heterogeneous MPSoC platforms. In this paper, we propose a novel unified design-mapping approach for efficient processing of vision GNN workloads on heterogeneous MPSoC platforms. Particularly, we develop MaGNAS, a mapping-aware Graph Neural Architecture Search framework. MaGNAS proposes a GNN architectural design space coupled with prospective mapping options on a heterogeneous SoC to identify model architectures that maximize on-device resource efficiency. To achieve this, MaGNAS employs a two-tier evolutionary search to identify optimal GNNs and mapping pairings that yield the best performance trade-offs. Through designing a supernet derived from the recent Vision GNN (ViG) architecture, we conducted experiments on four (04) state-of-the-art vision datasets using both ( i ) a real hardware SoC platform (NVIDIA Xavier AGX) and ( ii ) a performance/cost model simulator for DNN accelerators. Our experimental results demonstrate that MaGNAS is able to provide 1.57 × latency speedup and is 3.38 × more energy-efficient for several vision datasets executed on the Xavier MPSoC vs. the GPU-only deployment while sustaining an average 0.11% accuracy reduction from the baseline.