1 January 2025 in SSRN Electronic Journal
A blood group system used in medical and forensic biology to classify blood types based on the presence of specific molecules or antigens that are attached to red blood cells. The traditional detection methods for blood grouping are an invasive method and manual analysis, for which the results are too slow to obtain bringing incorrect diagnosis many times. In recent times AI & image processing techniques has proven to be a reliable way for automatic classification of the blood group. To address this issue, high-resolution images of the blood samples are acquired using a dedicated imaging system in this work. The pre-processing would include denoising, contrast stretching and image segmentation to improve the image quality and concentrate on regions of interest in blood cells. Then feature extraction methods used to discover distinctive features which could be shapes, color and texture or any other visual characteristic that differentiate the blood groups. The segregation features are used to train a deep neural network (DNN) for the classification of blood groups. Optimization Using Firefly Algorithm: To solve the performance issue on the AI model, the function is optimized by a technique inspired by the firefly called Firefly Algorithm. This algorithm return some best model parameters, like network architecture, learning rate and regularization which increase the blood group prediction accuracy. It can be observed from the output results that while integrating AI, image processing and FA refine the Bgroup classification accuracy as well as efficiency at a great height.
1 January 2025 in SSRN Electronic Journal
Strawberries are commonly grown and have an economic impact, but they have various diseases that can lower yield and quality significantly. This work is based on the implementation of Convolutional Neural Networks (CNNs) for a system that can reliably differentiate diseases in strawberry crops. A collection of 370 photos was compiled dealing with fungus, and illnesses caused by bacteria and fungi such as leaf blight, fruit rot, powdery mildew, and grey mould. To address the issue of limited datasets, along with splitting training and testing data at a ratio of 80:20, data augmentation techniques were applied. The CNN model used in this study was implemented in MATLAB, and three well-known architectures (VGG-16, GoogLeNet, and ResNet-50) were utilized for classifying disease. The results showed how well the CNN models performed, with ResNet-50 consistently outperforming the rest in terms of training accuracy. Adding image enhancement techniques improved accuracy rates, reasserting the importance of background complexity reduction. The conclusion demonstrates the capability of CNNs for effective strawberry disease diagnosis, paving new paths for timely diagnosis and treatment. This study bears great relevance to the development of plant disease identification in various crops with applications in agriculture and paves the way for the establishment of automated disease management systems in the production of strawberries in order to increase this strawberry productivity.
30 June 2025 in Asian Journal of Interdisciplinary Research
Development of generic skills is a primary focus for educational policies and accreditation bodies, there has been insufficient exploration into engineering students' perceptions of these skills and their motivation to develop them. Typically, the promotion of such competencies relies on conventional methods rather than well-organized curriculum frameworks. In addition, the absence of a unified definition and interpretation of generic skills across various disciplines complicates curriculum design. This study seeks to evaluate the competencies of engineering students in generic skills by employing an innovative decision-making approach. The proposed method assesses students' skills within a structured framework by combining the Analytic Hierarchy Process (AHP) and the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS). AHP method is utilized to determine the weights of different competency criteria, whereas the TOPSIS method is employed to rank criteria according to their proficiency in these skills. The approach taken in this study demonstrates the efficacy of addressing decision-making challenges and underscores its practical utility and dependability in evaluating the development of generic skills. Finally, this study's findings determined the importance set of generic skills in the context of the Indian higher education system.
1 January 2025 in SSRN Electronic Journal
The field of genomics, however, recently underwent a reformation with the inventions of high-throughput technologies generating an enormous amount of genetic data. Further, gaining insight from such data is a challenging task because of the complex interaction of genes with proteins and other macro-molecules in a biological system. In line with the challenge above, this study endeavored to apply some machine learning and AI principles to the automation of genetic data analyses. More specifically, path rule algorithms represent a route to genome-wide functional annotation by an AI approach. Our strategy focuses on the prioritization of candidate genes based on functional similarities, whereas gene functions are predicted by deep learning approaches from genomic properties. GNNs model functional interactions between genes while CNNs study genomic sequences and gene expression profiles. Various benchmark datasets have been shown where the proposed approach significantly outperforms traditional approaches by a margin in state-of-the-art performance both in accuracy and computational efficiency. Our approach can also be generalized well to new datasets and can behave robustly on different types of genomic data. In all, these findings indicate that AI-based techniques, in conjunction with algorithms based on path rules, represent a potent tool for functional genomics, offering new insights into biological mechanisms which may also point to novel targets for therapy in a wide range of diseases.
1 January 2025 in SSRN Electronic Journal
Supply chain management is crucial in the retail industry to ensure timely delivery of products along with appropriate coordination between suppliers, manufacturers, distributors, and retailers. The research focuses on a unique optimization for fast… Hybrid ADAM and Adadelta Algorithm use the advantages of both adaptive learning rates and convergence speed. The new approach proposed Algorithm in its use by integrating the aforementioned Algorithms aims to improve the Optimization process by providing versatility in decision making enhancing the performance and efficiency of the Assign Itself. Retail is a dynamic industry, which requires a rapid and well-adaptive approach to discovering and optimizing new opportunities and this hybrid method is well suited to provide those capabilities. Blockchain technology in supply chain holds plenty of benefits like increased traceability, immutability, and transparency. Utilizing the immutability of blockchain technology, the solution aims to build a tamper-proof and auditable system capable of verifying the provenance of products, tracking inventory levels, and tracing the flow of goods across the supply chain. Implementing the Hybrid ADAM and Adadelta Algorithm fused with Blockchain Technology is a novel step to revolutionize supply chain management in the retail sector.