Government Engineering College, Idukki

Gas sensors play a critical role in ensuring safety, environmental monitoring, and industrial process control. In this study, carbon nanostructures synthesized from onion waste were employed to develop free-standing Carbon/TiO2 thin films through a hydrothermal method, offering an eco-friendly approach for advanced sensing applications. Grown free-standing Carbon/TiO2 composite film exhibits a unique morphology, featuring tube-like carbon structures decorated with broccoli-flower-shaped TiO2 formations, with an average flower diameter of 8 ± 0.2 µm. A chemiresistive sensor fabricated from carbon/TiO2 composite shows a high response of 339

Tumors, especially brain tumors-such as glioma, Meningioma, and pituitary tumors-scanned using Magnetic Image Resonance (MRI) shows difficulty in their classification and explainability due to their complex structure and irregular shapes, and the standard CNN may focus on the wrong features. Also, the native XAI approach “Grad-Cam” may work well with CNN, but when it comes to vision transformers, “GradCam” often misses out on the important features from the provided sample. To overcome this limitation, we propose our system, which uses multi-modal Eva02 and DCNN (Deformable Convolutional Neural Network) hybrids fused together using Cross Attention Fusion, and its explainability is measured by an adaptive mechanism which have two methods: custommade Cross Attention Gradient-Driven Multi-Head Attention Rollout CAD-GMAR, which proved to provide better results when the transformer dominates the output, and Grad-Cam which provides result when DCNN dominates the output. This framework was proven to be reliable, robust, and accurate even against the noise present in the image sample on which the framework was tested.

Seismic pounding between adjacent buildings remains a critical concern, particularly when base isolation is employed and near-fault (NF) ground motions dominate the response. This study develops a Bayesian probabilistic framework to determine minimum separation gaps for adjacent reinforced concrete buildings with fixed-base (FB) and base-isolated (BI) supports. Two Reinforced Concrete (RC) (10-storey and 8-storey) were analyzed in four adjacency configurations under 44 far-field (FF) and NF ground motions, scaled to PGAs from 0.1 to 1.2 g. More than 12,000 nonlinear time-history analyses with gaps ranging from 25 to 500 mm detected pounding via nonlinear gap elements. Pounding probabilities were quantified using Bayesian updating with Jeffreys prior and Monte Carlo bootstrapping to derive posterior medians, 95

The identification of bird species through automatic analysis of their vocalizations shows potential for a range of fields, including ecology, conservation monitoring, and studies on vocal behavior. In recent times, this has emerged as a heavily researched area, with numerous studies employing deep-learning models to categorize bird calls. However, the performance of these models is often limited by the small and imbalanced datasets available. Mel-spectrogram and Mel frequency cepstral coefficients (MFCCs) are extracted in the front end. We explore the effectiveness of convolutional networks and aggregated residual transformation network (ResNeXt) with a cutting-edge Channel Attention model using sequential aggregation strategy, which we denote as SeqCA-Net, for multi-label bird classification. Two variants of SeqCA-Net have been proposed, namely, SeqCA-MS-CNN and SeqCA-MS-ResNeXt, respectively. Our objective is to develop a reliable model capable of real-time identification of bird species through their calls, even when constrained by a limited volume of training data. Our findings show that this approach, particularly the SeqCA-MS-CNN model, achieved an F1 metric of 86.22

Nowadays, seismic hazards increase day by day; it is not possible to stop any hazard, but the buildings can be strengthened. This study developed a framework for evaluating seismic reliability by considering six RC building models through analysis. The models include regular as well as irregular buildings with shear walls, bracings, dampers, and base isolation. Initially, three seismic response parameters were evaluated, such as storey drift, storey displacement, and base shear for all models. Three multi criteria decision making (MCDM) consisting of the Best Worst Method (BWM), Analytic Hierarchy Process (AHP), and Fuzzy AHP, were employed to quantify the importance of these parameters. The probability of reliability of each model were evaluated with the help of MCDM tools and the Weighted Sum Method, the reliability and probability of reliability of each model were evaluated. Model 2, consisting of dampers, shear walls, and bracings, exhibited the highest seismic reliability and lateral stiffness. A regular building with base isolation was considered as Model 3, where an increase in flexibility and a decrease in reliability were obtained. Moderate seismic performance was obtained from the irregular C-shaped and I-shaped models. Base shear was considered the most important seismic response parameter, followed by storey drift and displacement for all three MCDM tools. For reinforced concrete (RC) structures, nonlinear seismic analysis combining MCDM tools facilitates reliable structural performance evaluation in the proposed framework.