AMAbdul Malik

Abdul Malik

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

Abdul Malik has a total of 13 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Decomposition and Stacked Meta-Learning for Short-Term Electric Vehicle Load Forecasting

    by Syed Sajid Ullah, Gang Li, Muhammad Zunair Zamir, Abdul Malik, Saeed Mian Qaisar, Mudassar Riaz

    Accurate short-term electric vehicle (EV) load forecasting is critical for grid stability and demand response management amid rapidly increasing EV penetration. However, existing statistical, ensemble, and deep learning models often rely on implicit sequence learning and struggle to jointly capture multi-scale temporal structure, nonlinear interactions, and cross-dataset heterogeneity in EV charging demand. This paper proposes a decomposition-driven hybrid forecasting framework that integrates Seasonal-Trend decomposition using Loess (STL) with component-specific learning and stacked ensemble fusion. The EV load signal is decomposed into trend, seasonal, and residual components, each modeled using an algorithm aligned with its statistical characteristics. Component-level predictions are then fused through a Gradient Boosting Regressor meta-learner to enhance robustness and generalization. Experimental evaluation on four real-world EV charging datasets (ACN, Palo Alto, Perth, and Paris) demonstrates consistent performance gains over representative statistical, ensemble, and deep learning baselines. On the ACN dataset, the proposed framework achieves MAE 0.614, RMSE 0.797, and MAPE 12.7%, corresponding to more than 90% and 88% reductions in RMSE and MAPE, respectively, relative to the strongest benchmark. These results highlight the effectiveness and scalability of the proposed approach for practical smart grid deployment.


  • LiDAR Fusion-based Optimized Path Selection for Autonomous Vehicles using the Verification Model

    by Muhammad Nawaz Khan, Sokjoon Lee, Mohsin Shah, Arshad Khan, Abdul Malik, Ajab Khan

    Autonomous vehicles use LiDAR sensors to navigate 3D maps of their surroundings, enabling improved decision-making and road control systems. It utilizes laser light to measure the speed, distance, shape, direction, and perturbation of any 3D object in a point cloud. Data on point clouds is in high resolution and used to identify different objects of regular or irregular shapes. It is also used in safe path planning, path optimization, and object avoidance routes. Although point cloud data is robust and beneficial for safe navigation, errors and inconsistencies in the interpretation lead to erroneous judgments. Processing wrong information potentially resulting in hazardous circumstances with severe consequences of casualties and road accidents. In contrast, each path planning and path optimization policy should be thoroughly tested and validated before being applied to real-world scenarios. To provide a safer and collision-free path and to verify optimized path policy, we have proposed a technique called ”LiDAR Fusion-based Optimized Path Selection (LOPS) for Autonomous Vehicles using the Verification Model.” The proposed method uses an object detection strategy and instructs autonomous vehicles to use optimized paths for safe travel and collision-free path selection. The proposed scheme is formally modeled and verified for defined states in UPPAAL, and each state with defined properties has been checked. LOPS has been evaluated with different parameters, including the jerking rate, which ranges from 1.0% to 0.06% in various densities of autonomous vehicles. We have also calculated the effect of optimization of the path on the distance covered in unit time, and it is found that 50 vehicles require 20% more time to cover 50 km than 10, 5, or 2 vehicles. The lane usage is calculated in unit time and shows that using optimization in the lane scenario, lane 2 is 5-19 times more used in 5 minutes. LOPS is evaluated with two datasets, CARLA and NuScenes for different parameters, including path length, path smoothness, object avoidance success rate, path tracking accuracy, and decision delay comparisons.

  • An Efficient Framework for Detecting Model Inversion Attacks in Connected Healthcare Systems

    by Abdul Malik, Saeed Mian Qaisar, Muhammad Zahid Khan, Muhammad Nawaz Khan, Theyazn H. H. Aldhyani, Rayan H. Alsisi

    The rapid adoption of artificial intelligence (AI) in healthcare systems and consumer electronics is enhancing clinical decision-making and reshaping patient care. However, these advancements also raise serious security and privacy concerns, particularly model inversion attacks (MIAs), which reconstruct sensitive patient data from trained AI models. Addressing MIAs is essential to ensure trustworthiness and effectiveness of AI-driven healthcare solutions. To address these challenges, a novel four-stage detection framework is proposed to safeguard healthcare AI models against MIAs. Firstly, the data preprocessing and normalization layer eliminates the input noise and stabilizes the incoming query patterns. Onward, the behavioral feature extraction is carried out. It seeks to detect adversarial intent by profiling the temporal patterns as well as the irregularities found in the query behavior. Afterward, statistical anomaly scoring is applied using Mahalanobis distance learning to quantify deviations in query behavior. Lastly, an ensemble-based decision module, built on support vector machine and random forest classifiers, is used for the identification of MIAs. It combines anomaly scores and utilizes a threshold-based decision to accurately separate benign and possible MIA sessions while enhancing robustness against sophisticated attacks. Extensive experimental results on the NIH ChestX-ray14 and MIT-BIH datasets show that the proposed framework improves best-performing benchmarks in terms of precision by 1.5%, recall by 2.8%, F1-score by 1.18%, and AUROC by 2.23%, while reducing overhead by 9.09%.


  • QFedPhish: A Quantum-Inspired Federated Learning Approach for Detecting AI-Generated Phishing Attacks

    by Abdul Malik, Saeed Mian Qaisar

    Advancements in artificial intelligence (AI) offer opportunities but have also contributed to the emergence of AIgenerated attacks, escalating the complexity and scale of modern cyber threats. These attacks generate serious concerns for the traditional security systems in maintaining accuracy, privacy, false-positive control, and resilience in modern distributed environments. To cope with these concerns, this paper proposes QFedPhish, a quantum-inspired federated learning approach designed to identify and detect AI-generated phishing attacks. A combination of real-world phishing logs with synthetically generated large language model-based phishing attempts is built as a representative dataset. This data is then transformed through a quantum-inspired feature encoder, where random Fourier features and tensor-network compression capture subtle stylometric and structural cues that distinguish phishing content from legitimate content. Next, each participating client locally trains a lightweight classifier on the resulting embeddings, enabling model adaptation to heterogeneous environments without exposing raw data. Finally, model updates are aggregated using federated learning enhanced with differential privacy and resilient aggregation mechanisms. It mitigates risks of data leakage and poisoning attacks. Experimental evaluation demonstrates that QFedPhish consistently outperforms the best benchmark models, achieving gains of 0.10% in accuracy, 0.33% in precision, 0.41% in F1-score, 0.61% in AUC, along with reducing communication overhead by 3.89% and improving convergence speed by 3.66%.