Authors - Rashmi Vipat, Priyank Doshi Abstract - Agriculture plays a vital role in ensuring food security, yet traditional crop selection and yield estimation practices often fail to account for complex interactions among soil, climatic, and environmental factors. Recent advances in machine learning (ML) have shown significant potential in addressing these challenges by enabling data-driven decision support for farmers. This paper presents a comprehensive review of machine learning–based crop recommendation and yield prediction techniques, focusing on their effectiveness in improving agricultural productivity and sustainability. The study analyzes various supervised and ensemble learning models applied to soil quality parameters such as nitrogen, phosphorus, potassium, pH, moisture, and climatic variables. Emphasis is placed on multimodal data integration, highlighting how the fusion of soil, weather, and remote sensing data enhances prediction accuracy. The review also discusses current limitations, including data scarcity, model generalization, and real-time deployment challenges, particularly in resource-con-strained farming environments. Finally, the paper identifies key research gaps and future directions toward developing robust, scalable, and intelligent agricultural decision-support systems.
Authors - Amit Kalita, Himashree Kalita, Manjit Kalita, Abhijit Chakraborty, Kalpita Dey, Prajukta Deb Abstract - The significance of M- Health platforms to promote health equity has reached critical levels as digitalization in the healthcare sector continues to grow post pandemic. M-Health platform utilization in developing countries like Bangladesh has unique challenges: inconsistent adoption of the digital healthcare system, thus leading to a suboptimal delivery of healthcare services to customers. Using blended models i.e., Expectation-Confirmation Model (ECM), UTAUT2, and the DeLone & McLean IS Success Model, with Training on Virtual Consultation Skills as the moderating variable, the study intends to examine the adoption intention of healthcare providers to continuously use M-Health Platforms for a myriad of services like virtual consultation, remote patient monitoring, electronic prescriptions, and e-health record keeping. This study used Partial Least Squares Structural Equation Modeling (PLS-SEM) to evaluate 898 responses. Social influence, relative advantage, regulatory clarity, digital literacy, trust in technology, and system quality, which collectively improve doctors’ satisfaction with virtual consultation platforms, were identified as important to the purpose of the study. The results offer concrete steps that healthcare providers, platform creators, and policymakers can take to build and improve a solid and dependable M-Health platform that encourages sustained partnership with physicians by alleviating resistance that physicians may have about M-Health platforms in comparable developing countries.
Authors - Yaram Srinivasa Reddy, Bairoju Sreelatha, Shankar Lingam. M Abstract - Knowledge from a resource-rich source domain is leveraged in traditional transfer learning to enhance classification in a relatively data-scarce target domain. However, the resulting target models often suffer from overfitting and limited generalization, which restricts their utility in noisy and resource-constrained environments such as remote sensing. To mitigate these limitations, this work introduces a nuclear norm–regularized teacher–student framework for hyperspectral scene classification. In particular, the student model is regularized with the nuclear norm to encourage low-rank parameter representations, improving robustness to ambient noise. Further, we introduce a relative reconstruction loss (RRL) metric to measure the robustness of the student model to environment noise. Trained on several benchmark datasets, the proposed student model attains up to 87.0% classification accuracy on the independent test sets of UC Merced and EuroSAT, while remaining substantially lighter than the teacher network. Further, relative reconstruction values are computed for different amounts of noise added in the input space; RRL saturate to values less than 1.0 for all the datasets, substantiating that the regularized student model is indeed robust. The competitive performance of the regularized student model compared to the teacher network, its lightweight design, together with RRL values less than one, suggest that the proposed student model can effectively be deployed in noisy and resource-constrained environments such as edge and fog devices.
Authors - Nagesh Sharma, Priyanka Yadav, Kavita Singh Abstract - An accurate determination of childhood malnutrition is necessary for preventive measures. This paper proposes a modified scoring scheme comprising two new elements: the Integrated Anthropometric Score (IAS) and the Hybrid Integrated Score (HIS). IAS uses six primary anthropometric measurements, such as BMI, MUAC, WHZ, WAZ, HAZ, and skinfold thickness, along with selected interaction terms that capture the non-linear connections between growth parameters. The weights are determined by regularized logistic regression, allowing the score to be transparent while still adapting to the statistical structure of the data set. To further stabilize the predictions, the HIS combines BAI, IAS, and a machine learning probability component to make the predictions robust in both synthetic and real-world samples. The models were developed using a synthetic dataset of 9,456 children and tested with five-fold cross-validation and a separate real-world dataset of 38 children. Interaction selection and regularization were performed to control noise sensitivity and avoid overfitting. The findings indicate that the IAS model outperforms BAI with its higher cross-validated accuracy (0.93) and strong performance on real data (0.95). The HIS stays consistent in accuracy across areas and indicates better generalization. The results suggest that by combining multidimensional anthropometric characteristics, interaction-aware modeling, and hybrid learning, a new, more adaptable, and clinically interpretable tool for predicting nutritional risk has been developed, surpassing traditional composite indices.
Authors - Mehak Mukesh Agrawal, Saumya Kumari, Gaganam H V S M Soma Sai, Ankit A. Bhurane Abstract - Most existing artificial intelligence (AI) based assistants are cloud-dependent and require constant internet connectivity. User data is sent to external servers for processing. While this data is often encrypted, it is prone to risks such as cloud security threats. Additionally, users need to be cautious not to share sensitive information. To overcome the aforementioned privacy and internet availability concerns, this paper proposes a completely offline, on-device, cross-device, and open-source system to ensure complete data privacy. The proposed system was tested with several datasets, including AI2 Reasoning Challenge, SQuAD 1.1, CoNLL 2003, GSM8K and StrategyQA to evaluate the closed-form question answering (QA), contextual understanding, named entity recognition, mathematical reasoning and truthfulness, respectively, and with five on-device large language models (LLMs), including Gemma3 1B, SmolLM 1.7B, Qwen2 1.5B, TinyLlama-1.1B, and Phi-2. The system achieved the highest score for closed-form accuracy of 1.0. Its performance on reasoning ranged from 0.01 to 0.23. Truthfulness scores ranged from 0.24 to 0.59. High F1 scores for named entity recognition ranged from 0.74 to 0.79, and contextual understanding scores ranged from 0.02 to 0.17 across the different LLMs. The average response time of the system on mobile and desktop devices was evaluated and observed to vary according to system capability and model size. The system allows users to choose between multiple wake words specific to the Indian context. The proposed system functions on limited RAM and in constrained resource environments.
Authors - Mahzuzah Afrin, Rajasree Das Chaiti, Gazi Tahsina Sharmin Jahin, M. M. Musharaf Hussain, Mohammad Shamsul Arefin Abstract - Reliable identification of pneumonia from chest radiographs plays a central role in supporting clinical decision-making and patient management. Although deep learning models have shown favourable results for automated diagnosis, most existing studies rely on fully supervised training and mainly evaluate performance using accuracy or ROC-AUC metrics. Such evaluations may fail to capture clinical decision reliability, particularly in imbalanced medical datasets. In this work, we examine the effectiveness of self-supervised learning (SSL) for chest X-ray pneumonia classification through a controlled empirical study. A contrastive pretraining strategy is used to learn image representations from unlabeled chest X-rays, followed by supervised linear evaluation. The SSL-pretrained model is compared with a fully supervised model trained from scratch under identical experimental conditions. Our experiments reveal that the supervised baseline attains a slightly higher ROC-AUC; however, this improvement comes at the cost of increased false positive predictions, leading to lower overall accuracy. In contrast, the SSL-pretrained model exhibits a distinct prediction pattern. It achieves higher accuracy and notably improved precision and F1-score, indicating more balanced and reliable predictions. Precision– recall analysis further demonstrates the advantage of SSL in reducing false positive decisions. In addition, Grad-CAM visualizations suggest that the SSL-pretrained model focuses on clinically relevant lung regions. From a clinical decision-making perspective, these results suggest that self-supervised learning offers tangible advantages for chest X-ray analysis when prediction reliability is prioritized. This distinction is especially relevant in clinical settings.
Authors - Prerna Agarwal, Pranav Shrivastava, Samya Ali, Sachit Dadwal, Shubh Om Yadav, Saquib Hussain, Kareena Tuli Abstract - Most existing artificial intelligence (AI) based assistants are cloud-dependent and require constant internet connectivity. User data is sent to external servers for processing. While this data is often encrypted, it is prone to risks such as cloud security threats. Additionally, users need to be cautious not to share sensitive information. To overcome the aforementioned privacy and internet availability concerns, this paper proposes a completely offline, on-device, cross-device, and opensource system to ensure complete data privacy. The proposed system was tested with several datasets, including AI2 Reasoning Challenge, SQuAD 1.1, CoNLL 2003, GSM8K and StrategyQA to evaluate the closed-form question answering (QA), contextual understanding, named entity recognition, mathematical reasoning and truthfulness, respectively, and with five on-device large language models (LLMs), including Gemma3 1B, SmolLM 1.7B, Qwen2 1.5B, TinyLlama-1.1B, and Phi-2. The system achieved the highest score for closed-form accuracy of 1.0. Its performance on reasoning ranged from 0.01 to 0.23. Truthfulness scores ranged from 0.24 to 0.59. High F1 scores for named entity recognition ranged from 0.74 to 0.79, and contextual understanding scores ranged from 0.02 to 0.17 across the different LLMs. The average response time of the system on mobile and desktop devices was evaluated and observed to vary according to system capability and model size. The system allows users to choose between multiple wake words specific to the Indian context. The proposed system functions on limited RAM and in constrained resource environments.
Authors - Roshani Tawale, Jayshri Todase, Manisha Bharati Abstract - Enforcement of helmet regulations and accurate vehicle identification remain essential components of intelligent traffic management systems. Conventional supervision approaches depend heavily on manual inspection, which is labor-intensive and unsuitable for continuous large-scale monitoring. This study presents an automated framework for helmet violation detection and number plate lo-calization using the YOLOv8 deep learning architecture [3]. The proposed system supports static image analysis, recorded video processing, and live-stream detection within a unified pipeline. Performance is assessed using precision, re-call, and mean Average Precision (mAP@50). Experimental findings demonstrate consistent detection reliability and validate the framework’s applicability for real-time traffic surveillance systems.
Authors - Gunjan Pareek, Rajiv Singh, Swati Nigam Abstract - This research examines the transfer learning deep learning models in multimodal human activity recognition based on wearable sensor data. Raw IMU signals are converted to Gramian Angular Field (GAF) images to improve the feature representation and tested on WISDM and PAMAP2 datasets of 18 activity classes. Five CNN models, namely VGG16, MobileNetV2, ResNet50, DenseNet121, and EfficientNetB0, are trained and evaluated in the same conditions and measured by classification accuracy, statistical significance, and computation efficiency. GAF representations are always better than raw signals. DenseNet121 and ResNet50 have 99% accuracy, VGG16 and MobileNetV2 perform competitively and EfficientNetB0 performs worse. Most of the differences in performance are statistically significant (p < 0.05).
Authors - Devang Rupesh Dalvi, Gaurav Suresh Malik, Abhishek Jairaj Kunder Abstract - Prompt engineering has emerged as an essential paradigm in leveraging desired behaviors from large language models (LLMs) without altering their parameters. Although the majority of the current literature has revolved around the introduction of novel prompt engineering strategies, there has been comparatively less emphasis on the contribution of the evaluation and optimization of prompts in concrete systems. In this paper, we offer a specialized review of prompt engineering from an evaluation/optimization centric viewpoint with a larger nod to conceptual developments and illumination rather than detailing the comparisons of approaches. Furthermore, we attempt to establish the concrete importance of prompt engineering via a real-life application, which resulted in improved performances in tasks through the process of prompt refinement and informal evaluations without the need to change the architecture and weights of the models. The paper will also introduce the deficiencies in prompt engineering in the realms of re-producibility, robustness, and the unavailability of standardized approaches in the aspect of concrete evaluations.
