St. Joseph Dental College

The study introduces cross-modal memory-enhanced fusion (CrossMF) as a unified transformer-based model which performs emotion recognition from speech and text data. CrossMF combines dynamic attention between modalities with a memory enhancement system for effective fusion between textual and acoustic information. The training of CrossMF involves a two-step tactic where the acoustic encoder receives clean audio inputs from the Toronto emotional speech set (TESS) for optimization, yet the text encoder and fusion module acquire training from audio-text pairs from the multimodal EmotionLines (MELD) dataset. The system allows emotion predictions from three combination types, including audio-only, text-only, and audio with text inputs, due to adjustable modality access. The evaluation process takes place across the three different input scenarios to show that the system performs well with high generalization capability in both laboratory-recorded and natural conversational conditions. Fusion between multiple sources only occurs when both inputs are available so that integrity can be maintained until integration occurs. The model reaches a maximum validation accuracy of 97.68% while demonstrating sustained high-test performance, which proves its effectiveness when operated in various conditions without depending on manually created features. The architecture also supports future extensions, allowing developers to easily incorporate ablation studies and adaptive training strategies for real-world emotion-aware systems.

This paper presents a low-power and high-stability 12T SRAM cell designed using 18nm FinFET technologies and compared with conventional 6T, 8T, and 10T architectures. The proposed design employs read-path isolation, stacked transistors, and a leakage-controlled sleep mechanism to minimize dynamic and static power while improving read/write stability. Simulations were carried out in Cadence Virtuoso using PTM BSIM-CMG models under varied PVT conditions (0.8–1.0 V, –25 °C to 50 °C). The results show significant enhancements in performance metrics, achieving read SNM of 205 mV, hold SNM of 245 mV, and read/write delays of 52ps and 61ps, respectively. The proposed design demonstrates up to 60% reduction in read power and 28% overall power savings compared with traditional SRAM cells. The layout area of 0.61 µm² corresponds to an array density of 1.64 Mbits/mm², only a 1.25× overhead relative to a 6T cell. Monte Carlo simulations with 200 samples confirmed less than ±10% variation in power, ensuring robust and energy-efficient operation. The proposed 12T FinFET SRAM is thus a strong candidate for next-generation low-power, high-density memory applications in advanced VLSI systems.

In this work, we explore and compare two different approaches for iris recognition using the CASIA-IrisV1 dataset. The first approach uses DenseNet201 with minimal preprocessing to extract unique and distinct features from grayscale images of the eye. The second approach uses a more traditional way: it first performs iris segmentation, normalization, and contrast enhancement, then extracts features using the same DenseNet201 model. Both approaches create 512-dimensional feature vectors and compare them to confirm identity; Approach 1 uses cosine similarity while Approach 2 uses cosine distance. The tests show that Approach 1 outperforms Approach 2, with an accuracy of 90.59% compared to 84.71% for Approach 2. These results show that even minimal preprocessing can be enough to capture valuable features for iris recognition when employing powerful deep learning models like DenseNet201. The traditional approach is still helpful, especially when control over preprocessing steps and interpretability are essential. All things considered, this study makes clear the true amount of preprocessing needed and how it affects iris recognition algorithms.

Oil spills in marine environments, whether accidental or intentional, have catastrophic ecological and economic consequences. Traditional oil spill detection methods, such as manual surveys and visual inspections, are labor intensive, error prone, and lack scalability to monitor large oceanic regions. With advancements in satellite remote sensing, particularly Synthetic Aperture Radar (SAR) imagery, automated detection methods using deep learning have shown promising results. This study presents an oil spill detection system leveraging SAR images and deep learning techniques, including UNet for segmentation and classification. The proposed system enhances accuracy by distinguishing oil spills from look-alike phenomena such as algae blooms, wind slicks, and calm seas. The model is trained on a labeled dataset and evaluated using key performance metrics such as precision, recall, and Intersection over Union (IoU). Experimental results indicate that the system achieves high segmentation accuracy, with an IoU of 61.90 percent at optimal training conditions. The findings suggest that integrating deep learning models with remote sensing data offers a scalable and efficient solution for real-time oil spill monitoring, reducing environmental and economic impacts.

Mineral trioxide aggregate (MTA) is known for good sealing and also for its biocompatibility. However, limitations such as extendedsetting time and potential leakage persist. Therefore, it is of interest to compare the sealing efficacy of titanium-doped MTA andconventional MTA in the repair of access perforations using in vitro dye penetration model fourty extracted humanmandibular premolars were used. Titanium-doped MTA showed significantly less dye penetration (mean 0.92 mm) compared to conventional MTA(mean 1.46 mm) (p = 0.001), indicating superior sealing efficacy. Thus, Titanium-doped MTA demonstrates better sealing capability thanconventional MTA and may serve as an improved material for perforation repair.

This article discusses contemporary strategies to deal with immersed boundary (IB) frameworks useful for analyzing flow–structure interaction in complex settings. It focuses on immense advancements in various fields: biology, oscillation of structures due to fluid flow, deformable materials, thermal processes, settling particles, multiphase systems, and sound propagation. The discussion also involves a review of techniques addressing moving boundary conditions at complex interfaces. Evaluating practical examples and theoretical challenges that have been addressed by these frameworks are another focus of the article. Important results highlight the integration of IB methods with adaptive mesh refinement and high-order accuracy techniques, which enormously improve computational efficiency and precision in modeling complex solid–fluid interactions. The article also describes the evolution of IB methodologies in tackling problems of energy harvesting, bio-inspiration propulsion, and thermal-fluid coupling, which extends IB methodologies broadly in many scientific and industrial areas. More importantly, by bringing together different insights and paradigms from across disciplines, the study highlights the emerging trends in IB methodologies towards solving some of the most intricate challenges within the technical and scientific domains.

The rapid growth of Internet of Things (IoT) devices has exposed weaknesses in network infrastructures, so efficient and effective cyberattack detection is absolutely important. The application of Quantum Machine Learning (QML), more especially Quantum Support Vector Machines (QSVM), to raise the detection accuracy of known cyberthreats in IoT environments is investigated in this work. To better manage the enormous dimensionality and complexity of IoT network traffic data than conventional approaches, QSVM uses quantum computing ideas including improved quantum feature encoding and kernel techniques. With an accuracy of 99.12% against 92.00%, along with improved precision, recall, and F1-score metrics, experimental data show that QSVM greatly beats the classical Support Vector Machine (SVM). An essential first step toward integrating quantum computing into practical cybersecurity applications, our results show the potential of quantum-enhanced classifiers to deliver strong, scalable, and highly accurate intrusion detection systems for IoT networks.

Agriculture plays a crucial role in contributing about 17% of the gross domestic production (GDP) in India. The zest of rural population in India relies on farming and associated activities for their income. The use of traditional tools for agricultural activities like plow, spade, sickle, thresher, etc., gives less yield. Even though a lot of developments related to mechanization and automation of activities related to agriculture and farming are introduced, most of them are suitable for large-scale agriculturalists and farmers. The large-scale farmers associated with mass production are equipped with machines for cropping, thereby making huge profits. There is a need to develop automated/semi-automated systems for medium-or small-scale agricultural activities which would be energy efficient, easy to use and maintain, and economical. The manufacturers of agricultural machineries introduce machines which are imported in design, expensive, and suitable for large-scale applications like tillers, tractors, etc. However, developing and utilizing machines for specific purposes have been a challenge. The seed-sowing machines assist farmers in utilizing time and money to sow seeds in the desired position. In this chapter, the design and development of an Arduino-based battery-operated seed-sowing machine is discussed. The machine uses Arduino-controlled servo motors for controlled spacing and sowing of seeds. The battery eliminates the need for electricity to power the device. The Arduino-controlled design provides farmers a great level of precision and control over the operation of the machine. The depth, speed of tiller blades, and other parameters can be adjusted for the optimal performance of different types of crops and soil. The machine was tested for sowing the different types of seeds at a speed of 75 rpm. The motor and the battery capacities are 250 W and 21 Ah, respectively, and the battery charge will last for 1.3 hours. Overall, the Arduino-controlled multi-purpose seed-sowing machine is a valuable addition to the modern agricultural sector that helps the farmers to increase productivity, efficiency, and profitability. The proposed design is simple to use; hence, even an unskilled farmer can handle it. The chapter also discusses a comprehensive analysis on semi- and fully automated seed-sowing machines for agricultural purposes. Some of the shortcomings of traditional machines and practices are also compiled in this chapter. The scope for further improvement in seed-sowing machines and features is highlighted in the chapter, which encourages researchers to carry forward this research.

This paper develops an IOT smart surveillance system, aimed at providing real-time monitoring and security solutions for agricultural applications. Unfortunately, many of the crops get damaged due to theft and in most cases do not survive due to weather changes. The project works through integration of Arduino with the different sensors such as that of soil moisture and DHT11 compression with object detection along with automation like rain shed mechanism to provide real-time security and environmental monitoring for these related fields. That is on the flip side, to reduce human interruption, make good on the resource, and keep the crops in good health which all contributes to modernizing and sustaining agriculture. Such protection includes a rain shed mechanism for shielding crops from unanticipated weather occasions and soil moisture sensors for optimizing water application. The DHT11 monitors temperature and humidity to get an optimum environment for crop growth. In the case of security, the object detection takes photos and sends them via mobile to farmers, making surveillance real time. Therefore, the entire system intends to minimize human intervention, optimally utilize resources, and ensure healthy security of crops; thus modernizing agricultural practices while making environmental issues sustainable.

Artificial intelligence (AI) is transforming precision agriculture by enhancing soil analysis, crop tracking, and fertilizer management. AI-powered systems enable real-time monitoring of crop health, soil nutrient levels, and environmental conditions, providing farmers with actionable insights. Machine learning algorithms integrated with remote sensing technologies, such as drones and satellites, help monitor crops, predict yields, and detect early stress signs. In soil research, AI combines data from IoT sensors, GIS, and soil samples to create detailed soil health maps and predict changes. Predictive analytics optimize nutrient management by recommending precise application of water, fertilizers, and amendments based on real-time crop and soil needs. Automation, decision support systems (DSS), and variable rate technology improve input efficiency, reducing waste and environmental impact. By leveraging AI in these areas, precision agriculture promotes sustainable farming, lowers costs, and increases productivity through improved data analysis, real-time monitoring, and automation.