SRShazia Riaz

Shazia Riaz

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

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


Shazia Riaz has a total of 23 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Distinguishing Human-Generated and AI-Generated Academic Writing: A Machine Learning Benchmark Study

    by Ali Raza, Mohib Ullah, Rafiullah Khan, Adeem Ali Anwar, Muhammad Inam Ul Haq, Shazia Riaz

    The rapid adoption of large language models (LLMs) such as ChatGPT has raised critical questions about authorship, originality, and integrity in academic writing. Unlike conventional plagiarism testing tools, AI-generated or AI-rephrased text can preserve the original meaning and context of the text while modifying the writing style, making it challenging to detect using standard similarity checks. This study addresses this challenge by creating a domain-specific corpus of postgraduate-level academic texts. The corpus contains 22,520 samples, equally divided between human-written text and AI-rephrased text. All samples were preprocessed and represented using two common techniques: TF-IDF and Word2Vec. The dataset was evaluated using well-known machine learning and deep learning models, including Logistic Regression, Support Vector Machines, Recurrent Neural Networks, and transformer-based models BERT and T5. The results show that linear and sequential models provide low baseline performance, with accuracy between 50-54%. While BERT significantly outperforms the other models, achieving 83% precision along with a high recall rate. Confusion matrix analysis further shows that traditional models tend to overpredict AI authorship, whereas BERT demonstrates strong reliability in distinguishing between human-written and AI-generated text. The results show that transformer-based models are more effective for authorship verification in academic settings. They also emphasize the trade-offs among interpretability, computational cost, and predictive performance. In general, this study offers some important recommendations for the creation of credible, transparent, and domain-sensitive AI detectors for academia.


  • Open Access

    Stacked LSTM Model for Contextual Correlation Detection Among Multiple Emotions

    by Aqsa Younas, Shazia Riaz, Saqib Ali, Rafiullah Khan, Mohib Ullah, Daehan Kwak

    Emotions are closely tied to human behavior and play a critical role in daily life. With the widespread use of social media, individuals frequently express their feelings online, making emotion extraction from social networks an active area of research. This has applications in domains such as patient emotion monitoring, emotional text-to-speech synthesis, and empathetic chatbots. While numerous studies have focused on single-emotion detection, limited work has addressed the simultaneous identification of multiple coexisting emotions. These emotions are often influenced by contextual factors, including past emotional states and social interactions. Leveraging such dependencies can significantly enhance emotion detection performance. This study proposes a fast, accurate, and robust method to predict future emotions based on current emotional states, while also identifying interrelated emotions expressed in social media posts. We annotate a public dataset with multi-label emotion tags and model contextual dependencies—both temporal and social—using a novel Stacked Long Short-Term Memory (LSTM) architecture. The model incorporates encoder-decoder layers to capture short- and long-term dependencies among emotions and employs a time-distributed dense layer to differentiate features associated with each emotion. Our experiments show significant improvements in capturing contextual correlations among multiple emotions. The proposed model achieved a micro F1-score of 0.50 on the GoEmotions dataset and demonstrated comparable performance to transformer-based models on the DepressionEmo dataset, while requiring fewer computational resources. These findings underscore the effectiveness and efficiency of our approach in modeling complex emotional dynamics across social networks.


  • Open Access

    Detecting malicious code variants using convolutional neural network (CNN) with transfer learning

    by Nazish Younas, Shazia Riaz, Saqib Ali, Rafiullah Khan, Farman Ali, Daehan Kwak

    Malware presents a significant threat to computer networks and devices that lack robust defense mechanisms, despite the widespread use of anti-malware solutions. The rapid growth of the Internet has led to an increase in malicious code attacks, making them one of the most critical challenges in network security. Accurate identification and classification of malware variants are crucial for preventing data theft, security breaches, and other cyber risks. However, existing malware detection methods are often inefficient or inaccurate. Prior research has explored converting malicious code into grayscale images, but these approaches are often computationally intensive, especially in binary form. To address these challenges, we propose the Malware Variants Detection System (MVDS), a novel technique that transforms malicious code into color images, enhancing malware detection capabilities compared to traditional methods. Our approach leverages the richer information in color images to achieve higher classification accuracy than grayscale-based methods. We further improve the detection process by employing transfer learning to automatically identify and classify malware images based on their distinctive features. Empirical results demonstrate that MVDS achieves 97.98% accuracy with high detection speed, highlighting its potential for practical implementation in strengthening network security.


  • Open Access

    A Survey on Privacy-Preserving Caching at Network Edge: Classification, Solutions, and Challenges

    by Xianzhi Zhang, Yipeng Zhou, Di Wu, Quan Z. Sheng, Shazia Riaz, Miao Hu, Linchang Xiao

    Caching content at the edge network is a popular and effective technique widely deployed to alleviate the burden of network backhaul, shorten service delay, and improve service quality. However, there has been some controversy over privacy violations in caching content at the edge network. On the one hand, the multi-access open edge network provides an ideal entrance or interface for external attackers to obtain private data from edge caches by extracting sensitive information. On the other hand, privacy can be infringed on by curious edge caching providers through caching trace analysis targeting the achievement of better caching performance or higher profits. Therefore, an in-depth understanding of privacy issues in edge caching networks is vital and indispensable for creating a privacy-preserving caching service at the edge network. In this article, we are among the first to fill this gap by examining privacy-preserving techniques for caching content at the edge network. First, we provide an introduction to the background of privacy-preserving edge caching. Next, we summarize the key privacy issues and present a taxonomy for caching at the edge network from the perspective of private information. Additionally, we conduct a retrospective review of the state-of-the-art countermeasures against privacy leakage from content caching at the edge network. Finally, we conclude the survey and envision challenges for future research.


  • Open Access

    Sensors Faults Classification and Faulty Signals Reconstruction Using Deep Learning

    by Nayab Fatima, Shazia Riaz, Saqib Ali, Rafiullah Khan, Mohib Ullah, Daehan Kwak

    Sensor fault classification and reconstruction frameworks are crucial for the stable, safe, and reliable operations of Structural Health Monitoring (SHM) systems. Existing literature addressing reliability and efficiency is confronted with several challenges; especially, lacking a combined framework addressing both issues of classification and reconstruction at the same time. To tackle these issues, this paper proposes a fault-tolerant mechanism that uses various combinations of Deep Learning (DL) techniques to ensure the effectiveness and reliability of SHM systems in a resource-efficient way. The proposed mechanism is an integrated framework consisting of two modules: the sensor faults classification module and the faulty signal reconstruction module. We develop integrated architectures of CNN and RNN to classify faulty signals and employ various architectures of LSTM models for faulty signal reconstruction. Both modules are tested on the benchmark Canton Tower dataset. We augment the dataset with faulty signals created through simulations for an accurate analysis. The sensor faults classification module is evaluated by utilizing precision, recall, F1-score, and accuracy; it achieves a maximum accuracy of 94%. Additionally, the root mean square error (RMSE) value for the faulty signals’ reconstruction stands at zero. The experimental results show that our proposed mechanism outperforms existing state-of-the-art techniques regarding sensor fault classification accuracy and the quality of reconstructed faulty signals.


  • Open Access

    Automated Segmentation to Make Hidden Trigger Backdoor Attacks Robust against Deep Neural Networks

    by Saqib Ali, Sana Ashraf, Muhammad Sohaib Yousaf, Shazia Riaz, Guojun Wang

    The successful outcomes of deep learning (DL) algorithms in diverse fields have prompted researchers to consider backdoor attacks on DL models to defend them in practical applications. Adversarial examples could deceive a safety-critical system, which could lead to hazardous situations. To cope with this, we suggested a segmentation technique that makes hidden trigger backdoor attacks more robust. The tiny trigger patterns are conventionally established by a series of parameters encompassing their DNN size, location, color, shape, and other defining attributes. From the original triggers, alternate triggers are generated to control the backdoor patterns by a third party in addition to their original designer, which can produce a higher success rate than the original triggers. However, the significant downside of these approaches is the lack of automation in the scene segmentation phase, which results in the poor optimization of the threat model. We developed a novel technique that automatically generates alternate triggers to increase the effectiveness of triggers. Image denoising is performed for this purpose, followed by scene segmentation techniques to make the poisoned classifier more robust. The experimental results demonstrated that our proposed technique achieved 99% to 100% accuracy and helped reduce the vulnerabilities of DL models by exposing their loopholes.


  • The Future Prospects of Adversarial Nets

    by Muhammad Sohaib Yousaf, Shazia Riaz, Saqib Ali, Shuhong Chen, Guojun Wang

  • Open Access

    Aspect Based Sentiment Analysis of Ridesharing Platform Reviews for Kansei Engineering

    by Saqib Ali, Guojun Wang, Shazia Riaz

    At present online reviews are becoming an important source for Kansei engineering of the services provided by ridesharing platforms. Kansei engineering deals with incorporating customer feedback and demands into product and service design. Thus, it is used as a tool for organizations to uplift their businesses by considering customer reviews and feedback. Customer reviews available on social media are in unstructured form; therefore, sentiment analysis is employed to extract customer's opinions in a systematic manner. In India-Pakistan, these reviews are mostly in Roman Urdu/Hindi and English, which are of great value for ridesharing platforms as a part of their Kansei engineering strategy. However, sentiment analysis cannot be performed directly on these reviews as they are mostly in Roman Urdu/Hindi. Therefore, the objective of this paper is to conduct aspect based sentiment analysis on these reviews after translating them into English for Kansei engineering of the service. Consequently, sentiment analysis is carried out to extract the most frequent features along with nouns and adjectives used by the customers to express their sentiments. We extracted prominent aspects of the service (i.e., `Driver', `Company', `Service', and `Ride') based on their highest frequencies using aspect based sentiment analysis. The customer sentiments are then clustered into these main aspects using unsupervised machine learning technique. Each aspect is further analyzed based on their polarity, which serves as an input for Kansei engineering of the service. As a result, it can facilitate ridesharing companies to enhance their businesses by improving services in accordance with customer demands.


  • Open Access

    Pediatric Hodgkin’s Lymphoma: 5-Year Experience at Single Center in Pakistan

    by Shazia Riaz, Safia Khan, Farhana Badar

    OBJECTIVES: To review patients’ characteristics and treatment outcome pattern of pediatric hodgkinlympoma in a well funded center in developing country MATERIALS & METHOD: This was retrospective review of data of 301 pediatric hodgkin lymphoma patients up to 18 years of age treated in Shaukat Khanum Memeorial Cancer Hospital & Research Center, Lahore, Pakistan, during 2009 to 2013. RESULTS: Total No. of patients was 301, 81% were male. The median age of studied data was 9 years; ranging from 2 to 18 years. B symptoms in the form of fever, weight loss and night sweats were found in 51%, 29% and 34%, respectively. Median stage was 3 with neck bulky disease in 28% and Medistinal widening in 18% of patients. Trend of histopathology was mixed cellualrity at 56.5%. Most common treatment regime was ChlVPP/ABVD (59.5%); 30.6% of these 301 patient also got radiotherapy. At the end of treatment complete response was reported in 76.5%; 11.3% of these were relapsed afterward. CONCLUSION: Our data shows excellent outcome of pediatric hodgkin lympoma and results are comparable to that of developed countries.