Nadeem Iqbal Kajla
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
Nadeem Iqbal Kajla has a total of 42 co-authors. Here are the top co-authors by number of shared publications:
- Malik Daler Ali Awan
The Islamia University of Bahawalpur
5 shared publications - Malik Muhammad Saad Missen
Department of Software Engineering, The Islamia University of Bahawalpur, Bahawalpur, Pakistan
5 shared publications - Mickaël Coustaty
La Rochelle University
5 shared publications - Muhammad Muzzamil Luqman
La Rochelle University
5 shared publications - Chongsheng Zhang
School of Computer and Information Engineering, Henan University, Kaifeng, China
3 shared publications - Gaojuan Fan
College of Computer and Information Engineering, Henan University, Kaifeng, China
3 shared publications
Latest Publications
- 8 October 2026
- Published by Springer Nature
- p. 448-463
BHCP-DOC: A Real-World Bilingual Dataset for Handwritten Clinical Prescription Detection and Recognition
- 8 October 2026
- Published by Springer Nature
- p. 448-463
BHCP-DOC: A Real-World Bilingual Dataset for Handwritten Clinical Prescription Detection and Recognition
- Open Access
- 5 October 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21 (10) , e0359302
EffAtt-LungNet: An attention-guided EfficientNet framework for three-class lung histopathology classification
This study presents EffAtt-LungNet, an attention-guided deep learning framework for three-class lung histopathology classification. The model combines an EfficientNetB3 backbone with the Convolutional Block Attention Module (CBAM) to refine features and improve discrimination among benign lung tissue, lung adenocarcinoma, and lung squamous cell carcinoma. Experiments used the publicly released LC25000 lung subset, which contains 15,000 images across three classes and was itself produced by augmenting a smaller collection of underlying histopathology images. We generated 10,500 additional fixed variants, creating a balanced working corpus of 25,500 images with training, validation, and test partitions of 15,300, 2,550, and 7,650 images. EffAtt-LungNet reached 99.86% accuracy, 99.86% precision, 99.86% recall, 99.86% F1 score, and 0.9997 AUC on the image-level test partition. Performance was stable across five random seeds (mean accuracy 99.86 ± 0.02 % ), and paired errors differed from those of the EfficientNetB3 backbone (McNemar’s test, χ 2 = 47.27 , p < 0.001). Because pre-augmentation source identifiers for the released LC25000 images were unavailable, independence at the earliest source-image level could not be confirmed. The results therefore represent internal tile-level benchmark performance and require validation on independent, non-augmented, multi-centre datasets before any translational or clinical claim can be made.
- 5 October 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21 (10) , e0359302
Open AccessEffAtt-LungNet: An attention-guided EfficientNet framework for three-class lung histopathology classification
This study presents EffAtt-LungNet, an attention-guided deep learning framework for three-class lung histopathology classification. The model combines an EfficientNetB3 backbone with the Convolutional Block Attention Module (CBAM) to refine features and improve discrimination among benign lung tissue, lung adenocarcinoma, and lung squamous cell carcinoma. Experiments used the publicly released LC25000 lung subset, which contains 15,000 images across three classes and was itself produced by augmenting a smaller collection of underlying histopathology images. We generated 10,500 additional fixed variants, creating a balanced working corpus of 25,500 images with training, validation, and test partitions of 15,300, 2,550, and 7,650 images. EffAtt-LungNet reached 99.86% accuracy, 99.86% precision, 99.86% recall, 99.86% F1 score, and 0.9997 AUC on the image-level test partition. Performance was stable across five random seeds (mean accuracy 99.86 ± 0.02 % ), and paired errors differed from those of the EfficientNetB3 backbone (McNemar’s test, χ 2 = 47.27 , p < 0.001). Because pre-augmentation source identifiers for the released LC25000 images were unavailable, independence at the earliest source-image level could not be confirmed. The results therefore represent internal tile-level benchmark performance and require validation on independent, non-augmented, multi-centre datasets before any translational or clinical claim can be made.
- 4 January 2026
- Published by Springer Nature in International Journal on Document Analysis and Recognition (IJDAR)
- Vol. 29 (2) , 483-493
Comic character recognition using graph embedding in graph representations
- 4 January 2026
- Published by Springer Nature in International Journal on Document Analysis and Recognition (IJDAR)
- Vol. 29 (2) , 483-493
Comic character recognition using graph embedding in graph representations
- 15 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (6) , 11847-11858
ATTA-FL-Lite: Lightweight Byzantine-Robust Federated Learning for Resource-Constrained Medical IoT Devices
Federated learning (FL) enables privacy-preserving analytics at the medical IoT (IoMT) edge but is vulnerable to model poisoning and distribution shift. We present ATTA-FL-Lite, a lightweight aggregation rule that admits a client update only when three tests are jointly satisfied: (i) scale conformity via a median–absolute–deviation (MAD) z-score, (ii) directional alignment via cosine similarity to a coordinate-wise median reference, and (iii) non-degradation via a validation-loss z-score computed on a small, centrally held clean set. If no update passes, a coordinate-wise median fallback is used. We provide sub-Gaussian tail bounds for the loss test and an expected one-step descent bound for L-smooth objectives under benign mean-alignment and an accepted-set composition assumption; the analysis does not require a positive cosine threshold. Experiments on MNIST, Fashion-MNIST, and PathMNIST, under IID and non-IID partitions with up to 40% adversaries across four attack families, show that ATTA-FL-Lite maintains accuracy representatively ≈ 0.78–0.98 across Tiny/Small/Medium CNNs, reliably filters magnitude/noise attacks, and remains competitive against sign-flip. Server runtime scales approximately linearly with the number of participating clients at fixed model and validation sizes. These results indicate that ATTA-FL-Lite offers practical robustness for FL in resource-constrained IoMT deployments without cryptographic overhead or trusted root data beyond a small validation set.
- 15 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (6) , 11847-11858
ATTA-FL-Lite: Lightweight Byzantine-Robust Federated Learning for Resource-Constrained Medical IoT Devices
Federated learning (FL) enables privacy-preserving analytics at the medical IoT (IoMT) edge but is vulnerable to model poisoning and distribution shift. We present ATTA-FL-Lite, a lightweight aggregation rule that admits a client update only when three tests are jointly satisfied: (i) scale conformity via a median–absolute–deviation (MAD) z-score, (ii) directional alignment via cosine similarity to a coordinate-wise median reference, and (iii) non-degradation via a validation-loss z-score computed on a small, centrally held clean set. If no update passes, a coordinate-wise median fallback is used. We provide sub-Gaussian tail bounds for the loss test and an expected one-step descent bound for L-smooth objectives under benign mean-alignment and an accepted-set composition assumption; the analysis does not require a positive cosine threshold. Experiments on MNIST, Fashion-MNIST, and PathMNIST, under IID and non-IID partitions with up to 40% adversaries across four attack families, show that ATTA-FL-Lite maintains accuracy representatively ≈ 0.78–0.98 across Tiny/Small/Medium CNNs, reliably filters magnitude/noise attacks, and remains competitive against sign-flip. Server runtime scales approximately linearly with the number of participating clients at fixed model and validation sizes. These results indicate that ATTA-FL-Lite offers practical robustness for FL in resource-constrained IoMT deployments without cryptographic overhead or trusted root data beyond a small validation set.
- 18 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (4) , 6105-6115
PoisonShield-FL-NIDS: A Robust Defense Against Poisoning Attacks in Federated Learning Intrusion Detection
Federated learning (FL) has emerged as a privacy-preserving paradigm for collaborative intrusion detection in networked environments. However, it remains vulnerable to poisoning attacks (PAs) wherein malicious clients can corrupt the global model through deceptive updates. To address this, we propose PoisonShield-FL-network intrusion detection systems (NIDS), a robust FL-based intrusion detection system that integrates client-side anomaly filtering with trust-aware aggregation to defend against poisoned contributions. Experimental evaluation under varying levels of adversarial influence demonstrates that PoisonShield-FL-NIDS achieves superior performance across key metrics, attaining 93% accuracy, 91% precision, 94% recall, and an AUC of 0.96, while maintaining a low robustness index $ \text {RI}~\lt 0.05 $ even with 30% compromised clients. Compared to baseline FL models such as FL-CNN and FedACNN, our framework demonstrates faster convergence and higher resilience with only a marginal increase in communication overhead.
- 18 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (4) , 6105-6115
PoisonShield-FL-NIDS: A Robust Defense Against Poisoning Attacks in Federated Learning Intrusion Detection
Federated learning (FL) has emerged as a privacy-preserving paradigm for collaborative intrusion detection in networked environments. However, it remains vulnerable to poisoning attacks (PAs) wherein malicious clients can corrupt the global model through deceptive updates. To address this, we propose PoisonShield-FL-network intrusion detection systems (NIDS), a robust FL-based intrusion detection system that integrates client-side anomaly filtering with trust-aware aggregation to defend against poisoned contributions. Experimental evaluation under varying levels of adversarial influence demonstrates that PoisonShield-FL-NIDS achieves superior performance across key metrics, attaining 93% accuracy, 91% precision, 94% recall, and an AUC of 0.96, while maintaining a low robustness index $ \text {RI}~\lt 0.05 $ even with 30% compromised clients. Compared to baseline FL models such as FL-CNN and FedACNN, our framework demonstrates faster convergence and higher resilience with only a marginal increase in communication overhead.
- 8 June 2025
- Published by Springer Nature
- p. 167-178
Fusion of GNN and GBDT Models for Graph and Node Classification
- 8 June 2025
- Published by Springer Nature
- p. 167-178
Fusion of GNN and GBDT Models for Graph and Node Classification
- 16 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 71 (2) , 4959-4971
Q-BLAISE: Quantum-Resilient Blockchain and AI-Enhanced Security Protocol for Smart Grid IoT
The evolution of Smart Grids, a cornerstone of the Internet of Things (IoT), has revolutionized the electricity energy sector by enabling efficient, scalable, and secure energy management. As critical components of Smart Grid infrastructure, smart meters facilitate real-time data exchange between end users and service providers. However, transmitting sensitive data within these networks poses significant security challenges, particularly in the face of emerging quantum computing threats. Existing lightweight authentication and key exchange (AKE) protocols often fail to provide identity anonymity and impose substantial computational overhead, rendering them unsuitable for resource-constrained devices like smart meters. This paper presents a novel secure communication architecture for the Edge computing-assisted consumer devices and IoT (EACI) ecosystem, integrating Post-Quantum Cryptography (PQC) to ensure robust data integrity and access control against classical and quantum-era threats. The architecture employs a lightweight consortium blockchain, implemented using the Hyperledger Fabric framework, to maintain tamper-proof records of authentication events and energy transactions. Regional Edge Gateways (REGs) preprocess data and perform localized cryptographic operations, minimizing computational demands on resource-constrained devices. Furthermore, an AI-driven Intrusion Detection System (IDS) enhances the framework’s resilience by proactively detecting and mitigating security threats in real-time. The proposed architecture undergoes rigorous formal security analysis, demonstrating its robustness against quantum and classical cyber threats. Performance evaluations reveal a reduction of 31% in computational overhead, a 45% decrease in communication overhead, and a 15% improvement in energy efficiency compared to existing lightweight protocols, underscoring its suitability for resource-limited environments like smart meters. These findings establish the proposed architecture as a scalable, secure, and efficient solution for Smart Grids and other IoT environments, ensuring long-term data protection, system reliability, and operational sustainability in the quantum computing era.
- 16 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 71 (2) , 4959-4971
Q-BLAISE: Quantum-Resilient Blockchain and AI-Enhanced Security Protocol for Smart Grid IoT
The evolution of Smart Grids, a cornerstone of the Internet of Things (IoT), has revolutionized the electricity energy sector by enabling efficient, scalable, and secure energy management. As critical components of Smart Grid infrastructure, smart meters facilitate real-time data exchange between end users and service providers. However, transmitting sensitive data within these networks poses significant security challenges, particularly in the face of emerging quantum computing threats. Existing lightweight authentication and key exchange (AKE) protocols often fail to provide identity anonymity and impose substantial computational overhead, rendering them unsuitable for resource-constrained devices like smart meters. This paper presents a novel secure communication architecture for the Edge computing-assisted consumer devices and IoT (EACI) ecosystem, integrating Post-Quantum Cryptography (PQC) to ensure robust data integrity and access control against classical and quantum-era threats. The architecture employs a lightweight consortium blockchain, implemented using the Hyperledger Fabric framework, to maintain tamper-proof records of authentication events and energy transactions. Regional Edge Gateways (REGs) preprocess data and perform localized cryptographic operations, minimizing computational demands on resource-constrained devices. Furthermore, an AI-driven Intrusion Detection System (IDS) enhances the framework’s resilience by proactively detecting and mitigating security threats in real-time. The proposed architecture undergoes rigorous formal security analysis, demonstrating its robustness against quantum and classical cyber threats. Performance evaluations reveal a reduction of 31% in computational overhead, a 45% decrease in communication overhead, and a 15% improvement in energy efficiency compared to existing lightweight protocols, underscoring its suitability for resource-limited environments like smart meters. These findings establish the proposed architecture as a scalable, secure, and efficient solution for Smart Grids and other IoT environments, ensuring long-term data protection, system reliability, and operational sustainability in the quantum computing era.
- 5 November 2024
- Published by Elsevier in Pattern Recognition Letters
- Vol. 186, 286-291
A histogram-based approach to calculate graph similarity using graph neural networks
- 5 November 2024
- Published by Elsevier in Pattern Recognition Letters
- Vol. 186, 286-291
A histogram-based approach to calculate graph similarity using graph neural networks
- Open Access
- 1 August 2024
- Published by OMNISCHOLARS in Machines and Algorithms
- Vol. 3 (2) , 107-113
Rule-Based Capitalization Algorithm Using NLP for Text Formatting Consistency
Capitalization is essential in making texts readable, well-structured, and meaningful. This paper discusses the creation of a rule-based capitalization algorithm based on Natural Language Processing (NLP) methods for improving text formatting consistency. A representative dataset including news headlines, academic articles, and book titles is collected to achieve generalizability across text domains. The preprocessing step entails tokenization and part-of-speech (POS) tagging for categorizing words into notional (e.g., nouns, verbs, adjectives) and non-notional categories (e.g., articles, conjunctions, and brief prepositions). This categorization forms the basis for the structured capitalization rule application. The algorithm suggested has a systematic pipeline of tokenization, POS tagging, application of capitalization rules, and reconstruction of text. Executed with Python and NLP libraries like NLTK and spaCy, the algorithm capitalizes all notional words following title case conventions while maintaining linguistic and structural precision. The effectiveness of the algorithm is measured against manually formatted title case text and compared with available online converters for benchmarking. Precision, recall, and F1-score are used as performance metrics to measure accuracy and efficiency, and high reliability was shown in capitalizing text with few errors. A confusion matrix is utilized to examine classification accuracy, grouping outputs into true positives, false positives, false negatives, and true negatives. A Random Forest Model is also utilized to measure feature importance, with text reconstruction and exception handling emerging as central drivers of capitalization accuracy. The findings demonstrate that optimizing these elements greatly improves algorithm performance. The contribution of this work is to NLP-based text processing in presenting a rule-based, structured approach to capitalization that has implications in automated publishing, text formatting, and standardizing content.
- 1 August 2024
- Published by OMNISCHOLARS in Machines and Algorithms
- Vol. 3 (2) , 107-113
Open AccessRule-Based Capitalization Algorithm Using NLP for Text Formatting Consistency
Capitalization is essential in making texts readable, well-structured, and meaningful. This paper discusses the creation of a rule-based capitalization algorithm based on Natural Language Processing (NLP) methods for improving text formatting consistency. A representative dataset including news headlines, academic articles, and book titles is collected to achieve generalizability across text domains. The preprocessing step entails tokenization and part-of-speech (POS) tagging for categorizing words into notional (e.g., nouns, verbs, adjectives) and non-notional categories (e.g., articles, conjunctions, and brief prepositions). This categorization forms the basis for the structured capitalization rule application. The algorithm suggested has a systematic pipeline of tokenization, POS tagging, application of capitalization rules, and reconstruction of text. Executed with Python and NLP libraries like NLTK and spaCy, the algorithm capitalizes all notional words following title case conventions while maintaining linguistic and structural precision. The effectiveness of the algorithm is measured against manually formatted title case text and compared with available online converters for benchmarking. Precision, recall, and F1-score are used as performance metrics to measure accuracy and efficiency, and high reliability was shown in capitalizing text with few errors. A confusion matrix is utilized to examine classification accuracy, grouping outputs into true positives, false positives, false negatives, and true negatives. A Random Forest Model is also utilized to measure feature importance, with text reconstruction and exception handling emerging as central drivers of capitalization accuracy. The findings demonstrate that optimizing these elements greatly improves algorithm performance. The contribution of this work is to NLP-based text processing in presenting a rule-based, structured approach to capitalization that has implications in automated publishing, text formatting, and standardizing content.