Human–elephant conflict (HEC) remains a serious ecological and societal encounter, resulting in loss of human lives, crop destruction, and elephant mortality. Early detection of elephants and an accurate understanding of their behavioral state are essential for effective conflict mitigation. This work proposes a hybrid deep learning scheme for elephant detection (ED) and State-of-Mind (SoM) recognition, integrating handcrafted acoustic features and deep visual–spectral representations. Two complementary approaches are employed: (i) conventional audio feature extraction using Mel-frequency cepstral coefficients (MFCC), spectrogram, spectral centroid, and roll-off features followed by preprocessing and deep neural network (DNN) classification, and (ii) mel-spectrogram image representation processed using a pre-trained Visual Geometry Group-16 (VGG-16) network via transfer learning. To further enhance prevention, a Q-Learning–based decision-making module is introduced to dynamically trigger deterrent actions based on detected elephant presence and inferred SoM. The proposed system enables proactive, intelligent prevention of elephant–human encounters by combining perception, behavioral analysis, and reinforcement learning–based control. Experimental results demonstrate improved detection accuracy (99.3
ZnxCe1-xVO4 nanoparticles were prepared via a sol–gel auto-combustion method, providing a simple, economical, and eco-friendly approach. XRD confirmed the formation of a tetragonal zircon-type crystal structure with high phase purity, while FTIR revealed characteristic metal–oxygen bonding vibrations, validating the successful formation of the vanadate. FESEM coupled with EDX displayed homogeneous elemental distribution and uniform nanoparticle morphology, further supported by TEM and SAED, which confirmed nanoscale dimensions and polycrystalline nature. The CV profiles exhibited distinct redox peaks, confirming pseudocapacitive behaviour originating from reversible faradaic reactions. The tetragonal zircon-type crystal structure of ZnxCe1-x VO4 nanoparticles facilitates fast ion transport and reversible redox processes, enabling a high specific capacitance of 449 F g⁻1 at 1 A g⁻1, along with an energy density of 89.8 W h kg⁻1 at a power density of 599.77 W kg⁻1. The electrode exhibited remarkable cycling stability, maintaining 87.01
Traditional studies on mutations in the hemoglobin beta gene have relied solely on the statistical analysis of single-gene mutations, which do not adequately capture their sequential and context-dependent dynamics. Current models are limited in scalability, perform significantly worse across populations, and have limited biological interpretability. Some hurdles of deep learning include overfitting and poor integration in the genomic context. A hybrid framework, centered on higher-order Markov chains with fuzzy logic, is proposed to represent uncertain mutation states and to capture probabilistic dependencies for modeling temporal mutations. Statistical, alignment, structural, and dynamic analyses of 100 mutated hemoglobin beta sequences are performed using Clustal Omega, entropy measures, a hidden Markov model, and co-occurrence networks. Higher-order probabilistic Markov modeling with fuzzy automata performs superiorly in all evaluations. In splice prediction, Area Under Curve (AUC) scores of 0.67 (3′) and 0.82 (5′) are achieved, which outperformed existing models. High alignment quality (similarity percentage 0.98, column score 0.029, total consistency 700) with low run time (0.42 s) is achieved using Clustal Omega. The proposed method achieved an accuracy of 98.7
A common grounded voltage booster converter is introduced for fuel cell powered electric vehicles. The designed non-isolated structure employs an improved quadratic boost topology and a switched capacitor-based voltage lift network. It has a very high voltage boosting capability and provides ten times the applied input voltage at a very low duty ratio of 0.4. Beyond this, it has other key features such as low voltage stresses with a minimal component count, utilization of shared grounding between the source and load side, a continual source current that renders it ideal for interfacing the low voltage DC supply with the high voltage motor drive of electric vehicles. In addition, simultaneous switching makes the control circuit simpler. Working concepts of the designed topology are clearly elaborated in continuous, discontinuous, and boundary modes of operation. The effects of non-ideal components, power losses, and thermal stresses are also analyzed. Simulation outcomes are obtained from a MATLAB/Simulink platform. Theoretical analyses are validated through the built hardware model (240 V/200 W). Performance of the suggested structure is found to be superior when compared with various prevailing topologies.
Diabetic retinopathy (DR) is a long-term eye condition associated with diabetes, caused by prolonged hyperglycemia, that damages retinal blood vessels and may result in progressive vision impairment or blindness. In the early stages, DR often shows few or no clear symptoms, making timely detection clinically challenging. Accurate assessment of disease severity is therefore important for guiding appropriate clinical management and reducing the risk of vision loss. Conventional diagnosis relies on manual fundus examination by ophthalmologists. However, large-scale screening with these methods can be labor-intensive and time-consuming. To address these challenges, this research presents a deep learning framework to classify DR severity into five categories: no DR, mild, moderate, severe, and proliferative DR. The proposed method for detecting and grading DR relies on Convolutional Neural Networks (CNNs). The models were trained and evaluated using publicly available retinal fundus image datasets, with APTOS 2019 (Kaggle) serving as the primary dataset. For further evaluation and analysis, STARE, DRIVE, DIARETDB0, and CHASE were also utilized. Model performance was evaluated using five-fold stratified cross-validation to provide reliable estimates of classification performance. Among the evaluated architectures, InceptionResNetV2 demonstrated the best performance, achieving an overall accuracy of 93
