Saqib Ali's research while affiliated with University of Agriculture Faisalabad and other places

Trustworthiness and product traceability are essential factors in the apparel industry 4.0 for establishing successful business relationships among stakeholders such as customers, manufacturers, suppliers, and consumers. Each stakeholder has implemented different technology-based systems to record and track product transactions. However, these systems work in silos, and there is no intra-system communication, leading to a lack of complete supply chain traceability for all apparel stakeholders. Moreover, apparel stakeholders are reluctant to share their business information with business competitors; thus, they involve third-party auditors to ensure the quality of the final product. Furthermore, the apparel manufacturing industry faces challenges with counterfeit products, making it difficult for consumers to determine the authenticity of the products. Therefore, in this paper, a trustworthy apparel product traceability framework called ChainApparel is developed using the Internet of Things (IoT) and blockchain to address these challenges of authenticity and traceability of apparel products. Specifically, multiple smart contracts are designed and developed for registration, process execution, audit, fault, and product traceability to authorize, validate, and trace every business transaction among the apparel stakeholders. Further, the real-time performance analysis of ChainApparel is carried out regarding transaction throughput and latency by deploying the compute nodes at different geographical locations using Hyperledger Fabric. The results conclude that ChainApparel accomplished significant performance under diverse workloads while ensuring complete traceability along the complex supply chain of the apparel industry. Thus, the ChainApparel framework helps make the apparel product more trustworthy and transparent in the market while safeguarding trust among the industry stakeholders.

The extraordinary success of deep learning is made possible due to the availability of crowd-sourced large-scale training datasets. Mostly, these datasets contain personal and confidential information, thus, have great potential of being misused, raising privacy concerns. Consequently, privacy-preserving deep learning has become a primary research interest nowadays. One of the prominent approaches adopted to prevent the leakage of sensitive information about the training data is by implementing differential privacy during training for their differentially private training, which aims to preserve the privacy of deep learning models. Though these models are claimed to be a safeguard against privacy attacks targeting sensitive information, however, least amount of work is found in the literature to practically evaluate their capability by performing a sophisticated attack model on them. Recently, DP-BCD is proposed as an alternative to state-of-the-art DP-SGD, to preserve the privacy of deep-learning models, having low privacy cost and fast convergence speed with highly accurate prediction results. To check its practical capability, in this article, we analytically evaluate the impact of a sophisticated privacy attack called the membership inference attack against it in both black box as well as white box settings. More precisely, we inspect how much information can be inferred from a differentially private deep model’s training data. We evaluate our experiments on benchmark datasets using AUC, attacker advantage, precision, recall, and F1-score performance metrics. The experimental results exhibit that DP-BCD keeps its promise to preserve privacy against strong adversaries while providing acceptable model utility compared to state-of-the-art techniques.

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.

Short-term load forecasting (STLF) is part and parcel of the efficient working of power grid stations. Accurate forecasts help to detect the fault and enhance grid reliability for organizing sufficient energy transactions. STLF ranges from an hour ahead prediction to a day ahead prediction. Various electric load forecasting methods have been used in literature for electricity generation planning to meet future load demand. A perfect balance regarding generation and utilization is still lacking to avoid extra generation and misusage of electric load. Therefore, this paper utilizes Levenberg–Marquardt (LM) based Artificial Neural Network (ANN) technique to forecast the short-term electricity load for smart grids in a much better, more precise, and more accurate manner. For proper load forecasting, we take the most critical weather parameters along with historical load data in the form of time series grouped into seasons, i.e., winter and summer. Further, the presented model deals with each season’s load data by splitting it into weekdays and weekends. The historical load data of three years have been used to forecast week-ahead and day-ahead load demand after every thirty minutes making load forecast for a very short period. The proposed model is optimized using the Levenberg-Marquardt backpropagation algorithm to achieve results with comparable statistics. Mean Absolute Percent Error (MAPE), Root Mean Squared Error (RMSE), R², and R are used to evaluate the model. Compared with other recent machine learning-based mechanisms, our model presents the best experimental results with MAPE and R² scores of 1.3 and 0.99, respectively. The results prove that the proposed LM-based ANN model performs much better in accuracy and has the lowest error rates as compared to existing work.

Deep learning models have revolutionized AI tasks by producing accurate predictions. These models′ success largely depends on precise training using large-scale datasets, mostly crowdsourced from the target population. The training datasets may contain sensitive personal information, and the model parameters can encode this information on internal wirings of hidden layers, thus bearing the risk of a privacy breach. The modern trend of sharing trained models has grown the privacy breach risk manifold. The performance of existing privacy-preserving deep learning models trying to address this issue is unsatisfactory. Consequently, only a marginal proportion of these privacy-preserved models have been adopted in the industry. Therefore, we developed the first differentially private version of the block coordinate descent (BCD) algorithm. Our proposed mechanism considerably cuts the privacy cost by injecting an appropriate amount of noise in block variables. It achieves high accuracy comparable to its non-private equivalents. It improves the convergence speed and provides a provable privacy guarantee by performing privacy accounting using advanced composition and moments accountant methods. We empirically evaluate the robustness of our proposed mechanism on benchmark datasets. The results demonstrate the competitive performance in terms of both privacy cost reduction and speedy convergence against the state-of-the-art differential privacy-based mechanisms.

The preparation of Software Requirement Specification (SRS) document is a critical task as the successful completion of software depends heavely upon the SRS. The requirements gathering phase in Software Development Life Cycle SDLC is equally important as it witnesses the creation of an SRS document. The quality of an SRS document is dubious as there is a scarcity of proficient inspection methods for detecting the defects in the software. There are various reading based inspection techniques such as Checklist Based Reading technique (CBR), Defect Based Reading technique (DBR) and Perspective Based Reading technique (PBR). There are certain problems with these techniques such as the CBR only covers potential types of defects, The DBR focused on various defects but the problem is that it is not structured in nature. Likewise, PBR is not suitable as it mainly focused on use-cases. As owing to the absence of a suitable inspection technique, the main focus of this study is to highlight the Fagan based inspection technique. This technique is not only formal and lightweight but it also overcome the limitations of reading base d techniques. The Fagan is a structured approach and finds defects in specification, programming code, design and also others during the SDLC. During analysis a questionnaire based survey is conducted in almost 300 national and international software industries. As a result, 150 responses are received. The likert scale is used for developing the questionnaire. SPSS tool is also utilized for data entry and analysis. The consequences are drawn by developing hypothesis and using the Chi-square test. More satisfactory results are gained at international level than at the national level. The satisfaction level of defect detection is 59.2%, time saved is 48.3% and cost reduction is 40.0%. Thus, the obtained outcome is positive and satisfies the requirements.

There are five significant steps in software requirements engineering and the requirement validation (RV) step is one of them. Requirements validation ensures the reliability, stability, and consistency of the software. The persons involved in performing the requirements validation are stakeholders and external reviewers. The requirements validation techniques (RVTs) not only help in finding the errors but also help in guaranteeing the quality of the software. As quality is important, therefore, the main focus of this paper is to determine the ways of quality improvement in software requirements. A questionnaire-based survey is used to identify the most suitable RVTs for improving the quality. The results are compiled by developing a hypothesis and the Chi-square test method. The results showed that the cost reduction quality parameter is the most satisfactory because it reduces the cost up to 44.6% by using RVTs in Pakistan and foreign countries. Another important outcome is the level of effectiveness of RVTs which is much better in foreign countries as compared to Pakistan.

Feature extracting and training module can be done by using face recognition neural learning techniques. Moreover, these techniques are widely employed to extract features from human images. Some detection systems are capable to scan the full body, iris detection, and finger print detection systems. These systems have deployed for safety and security intension. In this research work, we compare different machine learning algorithms for face recognition. Four supervised face recognition machine-learning classifiers such as Principal Component Analysis (PCA), 1-nearest neighbor (1-NN), Linear Discriminant Analysis (LDA), and Support Vector Machine (SVM) are considered. The efficiency of multiple classification systems is also demonstrated and tested in terms of their ability to identify a face correctly. Face Recognition is a technique to identify faces of people whose images are stored in some databases and available in the form of datasets. Extensive experiments conducted on these datasets. The comparative analysis clearly shows that which machine-learning algorithm is the best in terms of accuracy of image detection. Despite the fact, other identification methods are also very effective; face recognition has remained a major focus of research due to its non-meddling nature and being the easy method of personal identification for people. The findings of this work would be useful identification of a suitable machine-learning algorithm in order to achieve better face recognition accuracy.

Machine learning has obtained remarkable achievement in longstanding tasks in various domains of artificial intelligence. However, machine learning certainly has some security threats, such as adversarial examples that hamper the machine learning models from correctly classifying the data. The adversarial examples are minor perturbations in the actual inputs to detract the model from its original task. Adversarial Attacks and their defenses are found in parallel when it comes to the literature of machine learning adversaries. In this paper, we have tried to inspect the adversarial attack types and their defenses by comprehensively classifying different techniques.