Category Archives: Uncategorized

Authors: Konka Kishan, Thuppathi Krishna Sree, Ramagiri Nissy Jasmine, Prathikantam Rakshitha

 

 

Abstract: Deep reinforcement learning (DRL) has excelled in video games but remains vulnerable to adversarial attacks. The project unveils Orthogonal Adversarial DRL (OADRL) to improve robustness in both discrete and continuous action spaces. OADRL integrates orthogonal regularization to limit overfitting and adversarial training to enhance resilience. The method assess against standard DRL models, measuring reward stability, adversarial robustness, and generalization. The project presents the OADRL reduces sensitivity to perturbations while maintaining high performance. OADRL improves robustness, ensuring smoother policies and greater resistance to adversarial noise. The insight highlight its potential for real-world applications like robotics and autonomous systems.

DOI: http://doi.org/

 

 

Published by: vikaspatanker

Authors: Dionysius Makumba

Abstract: The Zambian education system has evolved over the years since independence. Zambia inherited it’s education system from Britain and has continued mostly in the same line of the British education system. The Zambian government has been working hard to improve on the education system and the education standards of it’s citizens. The performance of pupils is of great importance to government as well as the general citizenry. According to Nelson Mandela, “Education is the greatest weapon which you can use to change the world. “Thus, government has been doing everything possible to improve on the performance of pupils in schools. Measures such as free education policy by the United Party for National Development (UPND) government evidenced by overwhelming enrollments recorded in many schools for the past three years . Consequently, over enrollment has caused poor academic performance for example the number of pupils enrolled does not match with the infrastructure and number of teachers. Most teachers have gone to towns. Another cause is that most parents or guardians do not encourage their children to go to school because of their low educational attainment. They are the ones who send their children to do something else when they are suppose to be in school thereby contributing more to high rate of absenteeism. The nature of the district also is another cause to poor academic performance for instance, the road network is extremely bad which did not facilitate movement of standard officers to monitor schools. Vehicles break down on the way therefore schools that are far away from the district education board secretary’s (DEBS) office are often not monitored.

 

 

 

Published by: vikaspatanker

Authors: Vaishnavi Chevale, Dr.Santosh Gaikwad, Dr. A. A. Khan, Dr. R. S. Deshpande§

 

 

Abstract: Contextual Emotion Recognition (CER) is crucial for human-computer interaction, requiring an understanding of emotions from linguistic and visual cues. This paper explores the integration of Large Vision- Language Models (LVLMs) to improve CER accuracy. The proposed framework employs multimodal learning to capture contextual dependencies, reduce biases, and enhance generalization. Experimental results demon- strate superior performance in real-world scenarios, de- creasing ambiguity and increasing robustness compared to traditional methods.

DOI: http://doi.org/

 

 

Published by: vikaspatanker

Authors: Amit Saxena

Abstract: The oil and gas industry heavily relies on complex machinery for extraction, processing, and transportation. Unexpected failures lead to costly downtime, safety hazards, and operational inefficiencies. Traditional maintenance strategies, such as reactive and scheduled maintenance, often fail to prevent unforeseen breakdowns, resulting in excessive costs and productivity losses. As the industry moves towards digital transformation, Artificial Intelligence (AI) and Machine Learning (ML) have emerged as powerful tools to revolutionize predictive maintenance strategies. By leveraging advanced data analytics, AI and ML can identify early signs of equipment failure, optimize maintenance schedules, and reduce unplanned outages. This paper explores the integration of AI/ML-driven predictive maintenance in the oil and gas sector, highlighting various techniques such as supervised and unsupervised learning, deep learning, and reinforcement learning. Additionally, it examines real-world applications, case studies from leading industry players, and the benefits of AI-driven maintenance, including cost savings, enhanced safety, and regulatory compliance. Despite the promising potential, challenges such as data quality issues, high implementation costs, and cybersecurity risks remain significant obstacles. We discuss strategies to overcome these challenges and explore future research directions in improving AI explainability and scalability. The findings demonstrate that AI and ML-based predictive maintenance not only enhance asset reliability but also contribute to sustainability efforts and operational efficiency, ensuring the long-term competitiveness of oil and gas companies in an increasingly digitized world.

 

 

Published by: vikaspatanker

Authors: Vijay Bisen, Dr. N.K. Singh

Abstract: One of the key issues in ensuring uninterrupted service is the handover process — the transition of a mobile device's connection from one base station to another. Traditional handover mechanisms, while functional, often struggle with the dynamic and complex environments of modern networks. This has led to increasing interest in leveraging machine learning (ML) to optimize and automate the handover process, enhancing both performance and user experience. Orthogonal Frequency Division Multiplexing (OFDM) and Non-Orthogonal Multiple access (NOMA) have been the leading contenders for modern wireless networks. NOMA is a technique in which multiple users data is separated in the power domain. In the proposed approach , a machine learning based handover mechanism between OFDM and NOMA has been proposed based on channel conditions. The condition for switching or handover has been chosen as the BER of the system. A comparative analysis with existing work indicates that the proposed scheme outperforms the existing techniques in terms of SNR requirement thereby making the system more practically useful for fading channel conditions.

 

 

Published by: vikaspatanker

Authors: Vijay Bisen, Dr. N.K. Singh

Abstract: Automated handovers are critically important for maintaining the Quality of Service (QoS) in wireless networks, typically in mobile UE scenarios.. With increasing number of users and multimedia applications, bandwidth efficiency in cellular networks has become a critical aspect for system design. Bandwidth is a vital resource shared by wireless networks. Hence its in critical to enhance bandwidth efficiency. Orthogonal Frequency Division Multiplexing (OFDM) and Non-Orthogonal Multiple access (NOMA) have been the leading contenders for modern wireless networks. NOMA is a technique in which multiple users data is separated in the power domain. A typical wireless system generally has the capability of automatic fall back or handover. In such cases, there can be a switching from one of the technologies to another parallel or co-existing technology in case of changes in system parameters such as Bit Error Rate (BER) etc. This paper presents a review on existing machine learning based approaches for handover prediction in future generation wireless networks. The salient features of each of the approaches has been highlighted along with identifying potential research gaps, rendering insights into potential search avenues in the domain.

 

 

Published by: vikaspatanker

Authors: Prabhakar Tripathi, Amit Thakur

Abstract: Engaging in physical activity and exercise is essential for maintaining a healthy lifestyle and is a key factor in preventing and enhancing health. However, certain sports and physical activities may present an inherent risk of injury. Some intrinsic, extrinsic, mutable, non-mutable and initiating events may contribute as causes of injury in sports. This thematic review will provide an overview of the mechanisms that lead to sports injuries and the various elements that influence them. It will also explore the effects of sports injuries, how technology and innovation can be used to manage these risks and injuries, the significance of early risk analysis, and finally, future trends and directions in artificial intelligence research to reduce the risk of sports injuries and the strategies for managing them. By amalgamating the current state of knowledge within this field, the author aims to enhance our comprehension of the complex interplay and intricate relationship between the mechanisms of sports injuries and the prevention, management, and treatment of such injuries using emerging and evolving technologies. It's essential to emphasize and underscore that advanced technologies should be seen as a complement and augmenting the role of healthcare professionals rather than substituting them, given the recognized limitations of the current system and the imperative necessity for personalized and tailored treatments that can vary from one athlete to another.

 

 

Published by: vikaspatanker

Authors: Nilesh Shrivas, Amit Thakur

Abstract: The development of artificial intelligence (AI) based medical Internet of Things (IoT) technology plays a crucial role in making the collection and exchange of medical information more convenient. However, security, privacy, and efficiency issues during information exchange have become pressing challenges. While many scholars have proposed solutions based on AI and blockchain to address these issues, few have focused on the impact of the slow consensus algorithm of blockchain on the efficiency of information exchange. To improve the efficiency of information exchange, we propose an information exchange approach based on AI and DA Genabled blockchain, providing a secure and efficient environment for information exchange in the medical IoT. Additionally, to enhance the efficiency of information exchange in the medical IoT, a novel tip selection algorithm is introduced to reduce the time delay in reaching consensus, thereby enabling faster acquisition of trusted information via blockchain. Simulation results demonstrate that compared to methods based on traditional DAG-enabled blockchain, the approach proposed in this paper improves the efficiency of information exchange.

 

 

Published by: vikaspatanker

Authors: Francis Mumba

Abstract: Exploring The Difficulties And Prospects Brought With The Adoption Of Computer Studies In Public Learning Institutions: A Case Study Of Four Selected Public Day Secondary Schools In Luwingu District Of Northern Province, Zambia

DOI:

Published by: vikaspatanker

Authors: Vaishnavi Yelnare, Dr.Santosh Gaikwad, Dr. A. A. Khan, Dr. R. S. Deshpandes

Abstract: This research undertakes an in-depth exploration into the utilization of machine learning algorithms for the recognition and classification of sign languages commonly used by individuals within the deaf and mute communities. We evaluate different models, such as CNNs, LSTMs, and hybrid networks, for gesture recogni- tion, image processing, and sequence classification. Chal- lenges including lighting, occlusion, inter-user variability, and data scarcity are addressed. Experiments are con- ducted on real-world datasets like RWTH-BOSTON and American Sign Language (ASL) to benchmark model performance. Our study contributes a scalable, real-time framework for sign language recognition, which aids in bridging communication gaps for the hearing-impaired community.

 

 

Published by: vikaspatanker