Sharnbasva University

The digital transformation of academic ecosystems requires systems capable of addressing diverse cognitive behaviors. Traditional models lack personalized pathways, which reduces student engagement. In this chapter per the authors, a comprehensive adaptive framework is proposed to enhance academic outcomes. By integrating predictive analytics, the system dynamically assesses performance, identifies knowledge gaps, and recommends personalized resources. A hybrid approach combining Random Forest for static prediction and Long Short-Term Memory networks for sequential analysis captures structural and temporal data patterns. Experimental evaluations demonstrate significantly improved prediction accuracy, faster dropout risk identification, and higher engagement. This architecture provides a robust, data-driven foundation for developing student-centric environments while democratizing intelligence for scalable institutional deployment.

This paper investigates the application of a hybrid deep learning model for analyzing human behavior from handwritten data, focusing on prediction accuracy, recall, F1-score, and privacy preservation. The proposed hybrid model integrates Convolutional Neural Networks (CNNs) for feature extraction, Recurrent Neural Networks (RNNs) for sequence learning, and Autoencoders for dimensionality reduction. The model achieved 92.5% accuracy, surpassing the traditional centralized model (90.5%). Although the hybrid model showed slightly lower precision (86.5%), it maintained comparable recall (87.8%) and F1-score (88.4%). The most significant advantage of the hybrid model lies in its privacy preservation, with 95% privacy ensured through federated learning, making it a suitable choice for applications handling sensitive data. While the hybrid model required slightly more training time (15 minutes per round) compared to the centralized model (12 minutes per round), it strikes a balance between privacy protection and computational efficiency. The ablation study showed that reducing the number of local models from 5 to 3 resulted in a drop in accuracy to 86.5%, underscoring the importance of model configuration in federated learning. This study confirms that the hybrid deep learning model, while slightly less efficient in terms of time, offers notable improvements in privacy and predictive performance. The findings suggest that federated learning in conjunction with deep learning models is a promising solution for scenarios requiring high privacy protection and accurate behavioral predictions. Future research will focus on optimizing the hybrid model's training time, improving its scalability, and exploring transfer learning across different domains and datasets.

The flow and thermal characteristics of a viscous fluid undergoing a first-order chemical reaction, while either absorbing heat or generating heat internally, as it flows vertically through a channel are discussed in this paper. Assumed all fluid properties (viscosity, thermal conductivity, etc.) to depend on temperature and the channel walls are held at constant but different temperatures. In order to obtain solutions to the three-dimensional nonlinear coupled dimensionless equations arising from the problem, the author proposed an analytical method. The non dimensional coupled nonlinear ordinary differential equations are solved analytically by the method of perturbation scheme. Using MATLAB, the distributions of velocity and temperature are presented graphically, and their variations with respect to each parameter are studied numerically. It is observed that the variation of thermal properties and chemical reactions has a remarkable effect on the velocity and temperature distributions in the case of viscous flow through a vertical channel.

Image denoising aids in the accurate reconstruction of the image from the noisy observation through the accurate elimination of the noisy components. Despite the advantages leveraged by the conventional image-denoising methods, there exist certain drawbacks, including the manual parameter tuning requirement, local convergence with increased computational complexity, and overfitting problems. Hence, to mitigate such drawbacks, an effective Distributed Attention-enabled Imperialistic U-Net (DAIU-Net) model is proposed in this research. Specifically, the Hybrid Distributed Attention (HDA) enhanced the denoising performance by selectively focusing on the most informative features, thereby reducing the computational complexity and improving the denoising performance. Additionally, the incorporation of Residual Flow (ResFl) descriptor-based feature extraction enhances the image denoising performance and reduces the data complexity challenges. Moreover, the utilization of the Imperialistic Learning Optimization (ILO) effectively fine-tunes the hyperparameters of DAIU-Net to address the local convergence issues. Besides, the modified U-Net aids in the better generation of the relevant features responsible for accurate denoising. Nevertheless, the performance evaluation in terms of performance metrics visualized the superiority of DAIU-Net by attaining 54.99dB PSNR, 0.81 SSIM, and 52.21dB SDME for medical images using the Alzheimer's dataset.

Although intelligent Chatbots have been widely used in the tourism and hospitality sector, a frame of reference based on theory on the effect of intelligent Chatbots on the tourist experience is still limited. The study introduces the Chatbot-Mediated Tourist Experience (CMTE) Framework based on UTAUT2, the Smart Tourism Destinations model, and experiential quality theory. Findings of a systematic review of 78 peer-reviewed studies (2018–2026), based on the PRISMA methodology, confirm that personalisation depth, response empathy, multilingual capability and real-time contextual awareness are consistent factors that contribute to positive outcomes in both pre-trip, in-trip and post-trip situations.