Akhtar Khalil

Akhtar Khalil

Auto-generated profile, based on publicly available information

Akhtar Khalil

Faculty of Computer Information Science, Higher Colleges of Technology, Abu Dhabi, United Arab Emirates

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
11 December 2025 in IEEE Access
Verified
Journal Article
Open access
MBMohammad Fazail Bangash
Mohammad Fazail Bangash
AAArsalan Arif
Arsalan Arif
MHMuhammad Hanif
Muhammad Hanif
AKAkhtar Khalil
Akhtar Khalil
AIAbid Imran
Abid Imran
Induction motors (IMs) are vital to industrial operations but are prone to electrical and mechanical faults such as bearing damage, stator winding issues, and broken rotor bars. Conventional fault diagnosis methods often rely on single-source data, which limits detection accuracy under dynamic conditions. This study presents an AI-based multi-source fault diagnosis approach using data from vibration, current, sound, and thermal sensors collected under varied load and speed conditions. Raw data was preprocessed, and features were extracted across time, frequency, and time-frequency domains for machine learning (ML) classification. For deep learning (DL), the sensor data was transformed into scalograms and fused with thermal images, which were then used to train a 2D convolutional neural network (CNN). ML models achieved up to 96% accuracy on individual sensors but showed reduced performance with data fusion. In contrast, the CNN with fused data achieved 99.2% accuracy. These results confirm the effectiveness of sensor fusion and deep learning in enhancing fault detection reliability and suggest a promising direction for intelligent predictive maintenance systems.

24 November 2025 in IEEE Access
Verified
Journal Article
Open access
FKFatima Khalid
Fatima Khalid
UHUsman Haider
Usman Haider
MHMuhammad Hanif
Muhammad Hanif
ARAhmar Rashid
Ahmar Rashid
AKAkhtar Khalil
Akhtar Khalil
Deepfake detection remains a pressing challenge due to the rapid evolution of forgery techniques and the demand for robust, generalizable, and interpretable solutions. We present a sparse prototype-based framework for deepfake detection that integrates class-specific dictionary learning with high-capacity face embeddings from the InceptionResNetV2 backbone. Discriminative dictionaries for real and manipulated faces are learned, and Orthogonal Matching Pursuit is employed to encode sparse representations. Classification decisions are derived from reconstruction residuals and sparse coefficient patterns, enabling transparent, traceable predictions. The proposed method is evaluated on Celeb-DF and five FaceForensics++ manipulation subsets, including FaceShifter, NeuralTexture, DeepFakes, FaceSwap, and Face2Face. It achieves up to 98.21% accuracy on FaceSwap and 92.05% on Celeb-DF, demonstrating competitive performance with state-of-the-art deep architectures while maintaining moderate computational cost. Beyond accuracy, the framework offers intrinsic interpretability by linking predictions to semantically meaningful prototypes, making it particularly suitable for forensic and high-stakes verification scenarios.

3 December 2025 in IEEE Access
Verified
Journal Article
Open access
DBDur-e-Zehra Baig
Dur-e-Zehra Baig
MHMuhammad Hanif
Muhammad Hanif
AKAkhtar Khalil
Akhtar Khalil
This study presents a noninvasive and cost-effective method for estimating oxygen consumption (V O2) during exercise using the Hammerstein system framework (HamS), relying on parameters such as heart rate (HR), respiratory rate (RpR) and exercise rate (ER), with ER reflecting the frequency of body movements during rhythmic exercises. The model requires prior knowledge of an individual’s maximum and resting V O2 for accurate predictions. Validation in two scenarios showed a mean quality of fit of 79.5% during cycling and 81% during rowing, demonstrating strong correlations between ER, RpR, HR, and V O2. Furthermore, a self-biofeedback system (SBF) achieved root mean square errors (RMSE) of 0.8236 ml/min/kg (SD = 0.6657 ml/min/kg) during rowing and 1.346 ml/min/kg (SD = 0.3745 ml / min / kg) during cycling, which confirms the model’s ability to provide accurate feedback in real time. These results underscore the potential of this approach for noninvasive exercise monitoring, offering a practical tool to optimize exercise prescriptions and improve safety and efficiency without the need for expensive and invasive equipment.