The rapid advancement of machine learning techniques has led to their widespread application in various domains, including water resources. However, snowmelt modeling remains an area that has not been extensively explored. In this study, we propose a state-of-the-art (SOTA) deep learning sequential model, leveraging a Temporal Convolutional Network (TCN), for snowmelt forecasting of the Hindu Kush Himalayan (HKH) region. To evaluate the performance of our proposed model, we conducted a comparative analysis with other popular models, including Support Vector Regression (SVR), Long Short-Term Memory (LSTM), and Transformer models. Furthermore, nested cross-validation (CV) was used with five outer folds and three inner folds, and hyperparameter tuning was performed on the inner folds. To evaluate the performance of the model, the Mean Absolute Error (MAE), Root-Mean-Square Error (RMSE), R square ($R^2$), Kling–Gupta Efficiency (KGE), and Nash–Sutcliffe Efficiency (NSE) were computed for each outer fold. The average metrics revealed that the TCN outperformed the other models, with an average MAE of 0.011, RMSE of 0.023, $R^2$ of 0.991, KGE of 0.992, and NSE of 0.991 for one-day forecasts of streamflow. The findings of this study demonstrate the effectiveness of the proposed deep learning model as compared to traditional machine learning approaches for snowmelt-driven streamflow forecasting. Moreover, the superior performance of this TCN highlights its potential as a promising deep learning model for similar hydrological applications.

This study explores Human Activity Recognition (HAR) using smartphone sensors to address the challenges posed by position-dependent datasets. We propose a position-independent system that leverages data from accelerometers, gyroscopes, linear accelerometers, and gravity sensors collected from smartphones placed either on the chest or in the left/right leg pocket. The performance of traditional machine learning algorithms (Decision Trees (DT), K-Nearest Neighbors (KNN), Random Forest (RF), Support Vector Classifier (SVC), and XGBoost) is compared against deep learning models (Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), Temporal Convolutional Networks (TCN), and Transformer models) under two sensor configurations. Our findings highlight that the Temporal Convolutional Network (TCN) model consistently outperforms other models, particularly in the four-sensor non-overlapping configuration, achieving the highest accuracy of 97.70%. Deep learning models such as LSTM, GRU, and Transformer also demonstrate strong performance, showcasing their effectiveness in capturing temporal dependencies in HAR tasks. Traditional machine learning models, including RF and XGBoost, provide reasonable performance but do not match the accuracy of deep learning models. Additionally, incorporating data from linear accelerometers and gravity sensors led to slight improvements over using accelerometer and gyroscope data alone. This research enhances the recognition of passenger behaviors for intelligent transportation systems, contributing to more efficient congestion management and emergency response strategies.

This paper investigates Electrocardiogram (ECG) rhythm classification using a progressive deep learning framework that combines time–frequency representations with complementary hand-crafted features. In the first stage, ECG signals from the PhysioNet Challenge 2017 dataset are transformed into scalograms and input to diverse architectures, including Simple Convolutional Neural Network (SimpleCNN), Residual Network with 18 Layers (ResNet-18), Convolutional Neural Network-Transformer (CNNTransformer), and Vision Transformer (ViT). ViT achieved the highest accuracy (0.8590) and F1-score (0.8524), demonstrating the feasibility of pure image-based ECG analysis, although scalograms alone showed variability across folds. In the second stage, scalograms were fused with scattering and statistical features, enhancing robustness and interpretability. FusionViT without dimensionality reduction achieved the best performance (accuracy = 0.8623, F1-score = 0.8528), while Fusion ResNet-18 offered a favorable trade-off between accuracy (0.8321) and inference efficiency (0.016 s per sample). The application of Principal Component Analysis (PCA) reduced the dimensionality of the feature from 509 to 27, reducing the computational cost while maintaining competitive performance (FusionViT precision = 0.8590). The results highlight a trade-off between efficiency and fine-grained temporal resolution. Training-time augmentations mitigated class imbalance, enabling lightweight inference (0.006–0.043 s per sample). For real-world use, the framework can run on wearable ECG devices or mobile health apps. Scalogram transformation and feature extraction occur on-device or at the edge, with efficient models like ResNet-18 enabling near real-time monitoring. Abnormal rhythm alerts can be sent instantly to users or clinicians. By combining visual and statistical signal features, optionally reduced with PCA, the framework achieves high accuracy, robustness, and efficiency for practical deployment.

The growing problem of riverine waste pollution threatens water sustainability in Nepal, which is driven largely by rapid urban development and inadequate waste management. This study presents the first comparative evaluation of object detection and segmentation models for automated waste detection using novel datasets of UAV-captured imagery. We introduce four datasets, two for object detection and two segmentation in Bishnumati and Bagmati river. We evaluated three training strategies—training from scratch, full-model fine-tuning, and fine-tuning with frozen layers—and assessed model generalizability through cross-river transfer learning. The study showed that fine-tuning pretrained weights is a better approach for training segmentation models, whereas freezing pretrained backbone layers is better for object detection. DeepLabv3+, trained by fine-tuning pretrained models, performed better than object detection models for waste detection, with precision and recall scores of 0.915 and 0.934 for Bagmati, and 0.913 and 0.939 for Bishnumati. Furthermore, transfer learning across locations improves object detection mAP scores and refines segmentation mask predictions, thereby uncovering previously undetected waste items. The transfer learning fine-tuned DeepLabv3+ model obtained an mIoU score of 0.849 for Bagmati to Bishnumati and 0.841 for Bishnumati to Bagmati transfer learning. Our novel datasets and our methodological blueprint for optimizing deep learning training strategies in riverine environments offer a scalable and adaptable pipeline for automated waste monitoring in riverine environments, particularly valuable for regions with limited surveying infrastructure and for geographically proximate rivers where transfer learning can be effectively applied.

The rapid advancement of machine learning techniques has led to their widespread application in various domains including water resources. However, snowmelt modeling remains an area that has not been extensively explored. In this study, we propose a state-of-the-art (SOTA) deep learning sequential model, leveraging the Temporal Convolutional Network (TCN), for snowmelt-driven discharge modeling in the Himalayan basin of the Hindu Kush Himalayan Region. To evaluate the performance of our proposed model, we conducted a comparative analysis with other popular models including Support Vector Regression (SVR), Long Short Term Memory (LSTM), and Transformer. Furthermore, Nested cross-validation (CV) is used with five outer folds and three inner folds, and hyper-parameter tuning is performed on the inner folds. To evaluate the performance of the model mean absolute error (MAE), root mean square error (RMSE), R square ($R^{2}$), Kling-Gupta Efficiency (KGE), and Nash-Sutcliffe Efficiency (NSE) are computed for each outer fold. The average metrics revealed that TCN outperformed the other models, with an average MAE of 0.011, RMSE of 0.023, $R^{2}$ of 0.991, KGE of 0.992, and NSE of 0.991. The findings of this study demonstrate the effectiveness of the deep learning model as compared to traditional machine learning approaches for snowmelt-driven streamflow forecasting. Moreover, the superior performance of TCN highlights its potential as a promising deep learning model for similar hydrological applications.