A Multi Model Deep Learning Framework for Sarcasm Detection in Regional Social Media: An Overview Anilkumar M. Bhadgale, M. Maheswari 2025 1st International Conference on Aiml Applications for Engineering and Technology Icaet 2025, 2025 Sarcasm detection in social media is a challenging task due to its inherent reliance on contextual cues, tone, and cultural nuances. In recent years, multi-model deep learning frameworks have emerged as a powerful approach for addressing these challenges, particularly in regional social media, where language variations and local idiomatic expressions complicate the detection process. This survey explores the latest developments in multi-model deep learning frameworks for sarcasm detection, focusing on their application in regional social media. The survey begins by reviewing foundational techniques in sarcasm detection, including traditional machine learning approaches that rely on handcrafted features. These methods, although effective in certain contexts, often fail to capture the subtleties of sarcasm in informal, region-specific languages. The advent of deep learning has led to significant advancements, particularly through models like Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Transformers. These architectures, combined with Natural Language Processing (NLP) techniques, have enhanced the ability to identify sarcasm through text analysis. However, single-modal approaches focusing solely on text fail to fully capture sarcasm's multimodal nature, especially on platforms where users often express themselves through a combination of text, images, emojis, and video. This has led to the development of multi-model frameworks that integrate various data modalities, such as text, image, and user behaviour, to better understand the context of sarcastic expressions. In regional social media, where local language and cultural symbols play a crucial role, these multi-model approaches prove even more valuable. This survey highlights key multi-model frameworks, emphasizing their use in regional settings. By examining datasets, model architectures, and evaluation metrics, the survey underscores the importance of combining textual and non-textual features to improve sarcasm detection accuracy. Furthermore, it discusses the challenges of applying these models to region-specific social media, including the scarcity of annotated data and the complexities of language diversity.
Blockchain-Assisted Puncturable Signcryption for Cloud and Fog-Hosted Industrial Cyber-Physical Systems Qingru Ma, Jian Shen, Vijayakumar Pandi, Brij B. Gupta, Varsha Arya, Maheswari Mahalinghum IEEE Network, 2025 The widespread adoption of emerging technologies, such as interconnected sensors and advanced automation systems, has led to rapid advancements in smart industrial environments. While industrial cyber-physical systems (ICPSs) bring efficiency and innovation to industrial processes, the security risks are also of great concern. The existing work provides many methods for data security and authentication for ICPSs. However, since the authentication status of ICPSs is changeable, the up-to-date methods can hardly meet the requirements for rapid authentication and secure data transmission. What’s more, large-scale message revocation incurs significant communication and computational costs. This paper proposes a blockchain-assisted identity-based puncturable signcryption (BC-IBPSC) scheme for cloud and fog-hosted ICPSs. First, based on the concepts of puncturable encryption and signature, a novel identity-based puncturable signcryption method is presented to achieve efficient verifiable data encryption. Second, the blockchain is utilized to record all valid and punctured messages, to achieve evidence storage and traceability. The results of the analysis indicate that the scheme ensures security and operates efficiently. In addition, experiments show that the size of the signcrypted files does not affect the value of Gas consumption on the blockchain when uploading proofs.
Explainable Federated Learning-Based Secure and Transparent Object Detection Model for Autonomous Vehicles Ganesh Gopal Devarajan, Thirunavukkarasan M, Maheswari Mahalinghum, Daniel Arockiam, Mu-En Wu, Rajendra Prasad Mahapatra IEEE Transactions on Consumer Electronics, 2025 Intelligent Transportation Systems (ITS) and autonomous vehicles (AVs) are two fast-moving research fields. Understanding the surroundings and acting quickly is essential for AV control, especially in dynamic environments. Object detection algorithms must function properly under these demanding circumstances. Even though deep learning models are excellent at detecting objects, they are sometimes criticized for having unclear decision-making. Recently, explainable artificial intelligence (XAI) has improved these judgments’ interpretability. This paper proposes a federated learning system that protects data privacy while enabling decentralized training across many AVs of an upgraded object detection algorithm, Yolov7-E. Grad-CAM visualization techniques are used to illustrate the deep learning models, and for easier interpretability, semantic data is fused with object detection. Enhanced security through federated learning, reducing data privacy risks by 40%. Based on comparative studies, it is demonstrated that the suggested model, which utilizes the federated learning paradigm, attains higher mAP and is appropriate for AV decision mechanisms while maintaining data privacy.
A Survey on Detection of Various Casting Defects Using Deep Learning Techniques M. Maheswari, N. C. Brintha 2nd International Conference on Intelligent Data Communication Technologies and Internet of Things Idciot 2024, 2024 In the mass production process, producing the product quality is a challenging task, because of the metal casting process, the presence of the product varies irregularly due to the kinds of defects. Identification of faulty products early in an automatic manner is one of the challenges in the industry. This work is a systematic review of various kinds of defects in the casting process, automated defect detection systems with deep learning approaches, and an analysis of their performance. Deep learning approaches are used to produce high-quality products in production lines and enhance the quality inspection process earlier in an automatic manner.
Analysis of Semiconductor Chip Performance using Medallion Architecture S. L. Jany Shabu, Nirmal Kumar.P, Rohith Kumar.S, M. Maheswari, J. Refonaa 7th International Conference on Inventive Computation Technologies Icict 2024, 2024 This research presents a meticulous investigation into semiconductor chip performance using a data engineering methodology. Employing the Medallion Architecture, the study comprehensively analyses various CPU performance metrics, including Integer Math, Floating Point Math, Random String Sorting, Data Encryption, Data Compression, Physics, Extended Instructions, and Single Thread operations. The project, consummated via Microsoft Fabric and Python integration, aims to decipher intricate CPU behaviours across diverse computational workloads. This in-depth exploration not only elucidates these metrics but also aims to derive critical insights essential for advancing chip design and optimizing computational efficiency. The findings of this rigorous analysis contribute significantly to the understanding and utilization of semiconductor chip performance within contemporary computing paradigms.
Analyzing Yelp Open-Source Data-Set in Azure Data-bricks Mugunth D, Maheshwaran R, Maheswari M, S. L. Jany Shabu, J. Refonaa 7th International Conference on Inventive Computation Technologies Icict 2024, 2024 A data-driven application, Yelp has always served and will always serve as one. It is one of the first companies that allows local businesses to receive reviews from their customers. Communities have always developed it in cooperation with each other. According to Yelp, the data-set contains information regarding businesses, reviews, users, and check-ins, and has continuously been updated since 2015. A project titled "Analyzing Yelp Open-Source Data-set in Azure Data Bricks" will analyze this data-set which has been made available as an open source for sentiment analysis and descriptive analysis. This study analyzes local business performance, business distribution, review ratings, and other factors, as well as check-in rates in American business locations over time. This analysis shows that the Yelp is losing reviews, tips, elite users, and check-ins.
YOLO Dietician Sirikonda Vaishnavi, Sanne Hruthika Shetty, M Maheswari, J.L. Jany Shabu, J. Refonaa Proceedings of the 3rd International Conference on Applied Artificial Intelligence and Computing Icaaic 2024, 2024 In recent years, the integration of Artificial Intelligence (AI) in healthcare has attracted a lot of attention, particularly in the domain of personalized nutrition. With the rising prevalence of diet-related health issues such as obesity, diabetes, and cardiovascular diseases, the primary objective of this research study is to design and implement a platform capable of analyzing individual preferences, nutritional needs, health goals, and lifestyle factors to offer customized dietary guidance and management. This research study proposes a AI Dietician by utilizing AI algorithms to provide personalized dietary recommendations. This study also covers the technical aspects and methods used in creating a AI dietician, including feature extraction, algorithm selection, data gathering strategies, and model training by ultimately contributing to the prevention of diet-related health issues and promoting long-term health outcomes.
Blockchain-Enabled Decentralized Trust Management and Secure Voting system Somisetty Reddaiah, Ramanabaoyina Subramanyam, J. Refonaa, S. L. Jany Shabu, Maheswari, S. Praveen Proceedings 2024 4th International Conference on Pervasive Computing and Social Networking Icpcsn 2024, 2024 Blockchain technology has seen a huge increase in popularity as a result of its ability to provide a transparent and decentralized platform for numerous applications. In order to solve the inadequacies of existing systems, a decentralized trust management and voting system that makes use of blockchain technology is proposed. The system uses the benefits of decentralization, transparency, and immutability of blockchain technology to ensure the security and integrity of voting and trust data. By employing smart contracts, the system automates the verification process and eliminates the need for middlemen, which reduces costs and improves efficiency. Decentralized consensus methods are also a part of the system, enabling reliable and impartial voting processes. The use of encryption safeguards the privacy and secrecy of the participants. The suggested system has the potential to revolutionize voting and trust management because it offers a secure and reliable environment.
CYBER WEB: An Extension for Scanning Websites K.S.V Sindhu, M. Maheswari, D. Ramalakshmi Vitecon 2023 2nd IEEE International Conference on Vision Towards Emerging Trends in Communication and Networking Technologies Proceedings, 2023
IoT based Smart Warehouse and Crop Monitoring System Jithina Jose, B Keerthi Samhitha, M. Maheswari, M. Selvi, Suja Cherukullapurath Mana Proceedings of the 5th International Conference on Trends in Electronics and Informatics Icoei 2021, 2021
