Due to the rapid development of cloud-based Cyber-Physical Systems (CPS) and smart hardware, a large volume of industrial data has been generated, posing challenges related to latency, scalability, and effective hardware-software integration. The paper provides a framework for an AI-enabled hardware-software co-design of real-time CPS data analytics. It is proposed that the smart sensor hardware, edge-assisted processing, and machine learning models are integrated into the proposed system to enable low-latency decision-making and optimal system performance. The co-design technique guarantees effective communication between hardware potential and software intelligence, reducing computational overhead and communication delays. The proposed framework is evaluated on 10,000 records, showing 91.3% prediction accuracy, 88.7% processing efficiency, 0.87% system stability, and a low computational latency of 198 ms. In addition, the general performance index is 0.89 and it demonstrates the balanced scaling, responsiveness and efficiency. The superiority demonstrated by comparative analysis over traditional, machine-learning-based, and hybrid models indicates that the proposed model is the best approach for real-time industrial analytics in dynamic CPS settings.
The ever-increasing volume of unstructured data in image and text format has posed a major challenge to conventional data analytics. The combination of deep learning and data analytics has become a promising solution to harness knowledge from such diverse and unstructured data sources. This work is a review of the integration of deep learning methods with data analytics platforms for efficient visual and textual knowledge extraction. Various state-of-the-art deep learning models such as convolutional neural networks (CNNs), recurrent neural networks (RNNs) and transformers are investigated for feature extraction, pattern matching and semantic interpretation in image and text mining applications. The integrated approach facilitates efficient, scalable, and accurate processing of multimodal big data. Use cases in various fields including medicine, social media, document understanding, security and recommendation systems are explored to showcase the impact of this approach. Moreover, challenges such as data diversity, interpretability, computational efficiency and ethical issues are discussed, alongside potential future work. This work emphasizes how the synergy between deep learning and data analytics enhances intelligent decision-making by transforming raw visual and textual data into actionable knowledge.
Government agencies worldwide are increasingly adopting artificial intelligence (AI) to improve efficiency, transparency, and decision-making. The Government of India’s new Gov.in Secure Intranet Platform exemplifies this trend by integrating agentic AI into everyday e-government operations. Unlike traditional rule-based systems, agentic AI systems can autonomously pursue goals, adapt to new information, and coordinate multi-step tasks[1][2]. This paper presents a detailed case study of Gov.in, highlighting its transition from legacy systems to intelligent AI-driven features. We describe the platform’s secure architecture and innovative capabilities—including an AI chatbot, a smart meeting scheduler (Gov AI Assistant), AI-generated minutes of meeting (MoM), and a Gov.in Secure Intranet-specific GPT model—and the underlying machine learning techniques (e.g. transformer-based NLP models like BERT, phonetic search, and summarization). We also discuss Gov.in Secure Intranet’s integration with services such as Bhashini (for multilingual translation/transliteration) and Swagatam (visitor appointment management). The paper analyzes benefits (such as 75% faster meeting scheduling[3]) and challenges (including bias, privacy, and regulatory constraints[4][5]) of deploying intelligent agents in government. By examining Gov.in Secure Intranet, we illustrate how agentic AI can transform public administration, and we consider implications for future AI governance and policy.
The rapid expansion of Internet of Things (IoT) deployments across diverse domains has intensified the demand for efficient and sustainable power management solutions. Traditional reliance on disposable batteries presents critical environmental and logistical limitations, necessitating novel alternatives. This research introduces a transformative framework that employs Unmanned Aerial Vehicles (UAVs) to assist a Green Base Station (GBS) equipped with wireless power transfer capabilities to provide on-demand charging for widespread IoT devices. Central to our approach is the integration of Clustering and Routing Approach, which enables UAVs to autonomously learn optimal routing and charging policies in real-time, adapting to dynamic energy demands and environmental conditions. To validate our approach, we conduct extensive simulations comparing the proposed method with conventional clustering-based techniques, including K-Means, Hierarchical, and Greedy algorithms. The results demonstrate superior performance in operational efficiency, scalability, and responsiveness. This study underscores the potential of intelligent, UAV-assisted wireless charging as a sustainable and adaptive energy solution, paving the way for more resilient and autonomous IoT infrastructures.
Nowadays, a major challenge is the increasing distortion of information on social media platforms like WhatsApp. Ensuring the authenticity of user-generated content has become critical to prevent the spread of misinformation. In this paper, the authors propose a hybrid model that integrates machine learning (ML), deep learning (DL), and knowledge graph techniques to classify WhatsApp messages as real or fake. The approach begins with natural language processing (NLP) based preprocessing, dataset balancing, and Term Frequency–Inverse Document Frequency (TF-IDF) feature extraction. For classification, multiple ML algorithms such as support vector machine (SVM), logistic regression, decision tree, and XGBoost are evaluated, along with DL models including deep neural networks (DNN), long short-term memory (LSTM), gated recurrent unit (GRU), and fine-tuned Bidirectional Encoder Representations from Transformers (BERT). To further enhance classification performance, a knowledge graph built from high-frequency phrases is used in conjunction with a graph-based contextual scoring system. The suggested hybrid approach outperforms solo ML or DL models in terms of accuracy, precision, recall, and F1-score, according to experimental results on a real-world WhatsApp dataset.
