University of Engineering & Management

The research investigates how various risk criteria impact Hybrid and ELSS fund performance metrics through its assessment of annualised returns and Sharpe's Ratio and Treynor's Ratio and Jensen's Alpha. This study evaluates fund performance through assessment of risk characteristics which include correlation and annualised daily variance and downside standard deviation and Risk (Systematic) and Risk (Unsystematic). The results demonstrate that both fund categories experience annualized return and Jensen's Alpha decrease because of downside risk, while Hybrid funds face more extensive Risk (Systematic) challenges. The analysis of Sharpe's Ratio shows that both fund categories depend on downside risk assessment, but ELSS funds demonstrate stronger effects from this risk. The analysis of Treynor's Ratio demonstrates that ELSS funds require both downside risk and variance assessment, as these factors provide better understanding of fund performance. The study shows that investors must examine risk parameters when assessing fund performance because risk parameters influence performance metrics differently across various fund categories.

This research paper says about how people with physical disabilities feel and think when they use assistive technology, like wheelchairs or talking devices, to help them in daily life. It shows that using these tools can make people feel more independent and confident, helping them take part in family, school, and work activities. But sometimes, people may feel shy or worried about what others think when using these devices. The paper explains how important it is for society to accept and support people who use assistive technology so they feel comfortable and happy. It also suggests designing these tools in a way that helps people feel good about themselves and making sure everyone understands how helpful these devices are. Helping users feel accepted and supported can make a big difference in their lives. this study examines the social acceptance and psychological impact of assistive technology on physically challenged individuals. Assistive technologies (AT) have the potential to significantly enhance the quality of life by promoting independence, inclusivity, and participation in educational and social activities. However, issues such as social stigma, inadequate training, and mismatches between user needs and device functionality affect their acceptance and use. Social perceptions and emotional experiences play crucial roles in the adoption of AT, with users often facing stigma and embarrassment. Yet, AT can also transform users' psychological experiences by fostering academic and social engagement. Understanding these dynamics requires a user-centered design approach that addresses individual and societal challenges, aiming to reduce stigma and enhance the overall efficiency and acceptance of assistive technologies. Future research and innovation should focus on creating devices that cater to the diverse needs of users while promoting social inclusion and reducing psychological barriers.

The digital interactions have undergone an exponential trend that has fundamentally shifted the customer relationship management (CRM) systems, and advanced technological interventions are required to handle the more intricate expectations and behavior of the customer. The conventional CRM strategies are unable to handle large volumes of unstructured customer information, forecast subtle behavioral patterns, and provide personalized experiences at a large scale to generate lower customer satisfaction and loyalty. This is a broad literature review on the impacts of machine learning (ML) and deep learning (DL) in revolutionizing CRM practices in increasing customer experience, generating sustainable value, engaging, and gaining long-term loyalty. This paper is a systematic review conducted by relying on the PRISMA methodology to gather and analyze the current research. As it has been shown in the review, ML and DL technologies such as natural language processing, recommendation systems, predictive analytics, and sentiment analysis began to have impressive advantages in terms of the accuracy of customer segmentation, the ability to predict churn, lifetime value estimation, and personalization strategies. More so, incorporation of generative AI, federated learning, and explainable AI models is the next innovation in the development of transparent, privacy and ethically sound CRM systems. The study reveals that there are important gaps in the existing literature about the concepts of real-time emotion recognition, cross-channel behavior synthesis and sustainable AI implementations, and proposes the opportunities in edge computing integration, quantum machine learning applications, and neuromorphic computing to optimize CRM.

Existence in a globalized economy presents the most significant challenges that the supply chain management has never experienced before, such as demand in variation, disruption in supply, transparency deficiencies, and issues of sustainability. This review explicitly looks at how machine learning, deep learning and blockchain technologies can be used to transform the activity in the supply chains in a revolutionizing way. The work applies the PRISMA methodology to analyze the current trends of the supply chain systems with respect to intelligent supply chain system systematically and within the framework of new trends and future directions. The machine learning algorithms have amazing performance in demand forecasting, inventory optimization and predictive maintenance, whereas deep learning architectures perform better in complicated pattern recognition, computer vision tasks and natural language processing to supply chain analytics. Distributed ledger systems, smart contracts, and decentralized consensus mechanisms are all innovations that blockchain technology suggests which will bring unprecedented levels of transparency, traceability, and security. According to this review, existing research has a number of gaps such as limited integration frameworks involving each of the three technologies, lack of confirmation of scalability in practice, and little attention to the problems of adoption in small and medium enterprises. The article shows that hybrid solutions that integrate machine learning predictive functionalities with blockchain trust infrastructures have a higher performance on providing end-to-end supply chain visibility and optimization. The potential opportunities are autonomous supply chain orchestration, circular economy enabling and robust network design.

Electroencephalography (EEG)-based control has emerged as a promising solution for intuitive and non-invasive control of assistive robotic devices, particularly for individuals with severe limb loss or neuromuscular impairments where muscle signals are unavailable. This paper presents the design and development of a low-cost, lightweight 3D-printed prosthetic arm controlled exclusively by EEG signals. The proposed system decodes user movement intention directly from brain activity using signal processing, sensor fusion, and adaptive machine learning techniques. EEG signals are acquired through a non—invasive multi-channel headset, conditioned to suppress artifacts, and transformed into meaningful features representing motor imagery and movement intention. To improve reliability and reduce noise sensitivity inherent in EEG signals, multiple signal fusion and filtering techniques—including averaging, weighted averaging, Principal Component Analysis (PCA), and Kalman filtering—are investigated. Experimental results demonstrate that Kalman filter-based fusion significantly enhances signal stability and control accuracy in real-time prosthetic operation. The prosthetic arm employs a modular 3D-printed mechanical structure with servo-driven actuation, ensuring affordability and customization. Additionally, a future bidirectional feedback framework based on deep learning is proposed to enhance user adaptation and sensory perception. This work contributes toward accessible, intelligent, and fully brain-controlled prosthetic systems for medical rehabilitation and assistive healthcare.

This paper provides a literature review, comparative analysis and presents an original research of the use of Artificial Intelligence (AI) and Machine Learning (ML) technologies for cybersecurity with particular attention to intrusion detection. First, the research looks at conventional ML methods to make an analysis on malicious network activity detection on the CICIDS 2017 dataset. On top of the foundations, a network based large language model phi-3-mini is fine-tuned for intrusion detection using structured and network flow data. It takes natural language inputs of network properties, and enables the ability to predict scenario various intrusion or benignity. The combination of structured queries and natural language processing also presents a new mode for interacting with security systems, particularly for expert users. We also review literature on techniques used in AI based threat detection and malware analysis as well as automated incident response. Although the results show the promise of large language models to improve detection performance and accessibility above the baseline of what already exist, the study discusses challenges such as interpretability, resource demands, and robustness against adversarial inputs. In the post-Quantum age, such a model would be much valuable for assisting human operators to reinforce traditional network perimeter defenses and other security systems, that makes the current generation (albeit somewhat minimally) more effective against increasingly sophisticated and dynamic blackhat techniques of various threats.

Today's criminal identification systems encounter major obstacles while matching textual descriptions to visual databases since they use manual processes that show poor scalability and human-dependent decision making. The research introduces an automatic multi-modal framework which combines CLIP for text-image semantic linking with FaceNet for biometric feature extraction and EfficientNet for visual attribute detection. The pipeline system receives natural language request input, then extracts meaningful weighted attributes (semantics at 70% and attributes at 20% with face information at 10%) before finding top-k matches through a combination scoring method. The system achieves an 94% hit rate appearing in the top 10 results against the database of suspects. The paper presents technical specifications about face detection using MTCNN alongside confidence threshold values (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\geq 0.9$</tex>) and attribute matching through Jaccard approach which are examined together with operational ethics related to real-world application.

Electrical power plays a significant role in people's daily activities. Due to the advancement of technology IoT technique is used in most real-time applications. Various sensors are used to collect real data from the environment or objects. The research paper proposes an IoT-based health monitoring system for transformers. This system aims to improve the reliability and efficiency of electrical power distribution by enabling real-time monitoring of critical parameters like temperature, voltage, and current of the oil that is stored inside the transformer to keep the transformer healthy.. This mobile system will help the utilities to optimally utilize transformers and identify problems before any catastrophic failure occurs. This system will be an advanced step to the automation by diminishing human dependency. As it is a wireless communicating system, there is no need of large cables which are of high cost. Thus, THMS offers a more improved transformer monitoring.