Search Results (47)

Human activity recognition (HAR) based on wearable sensors is a key technology in the fields of smart sensing and health monitoring. With the rapid development of deep learning, its powerful feature extraction capabilities have significantly enhanced recognition performance and reduced reliance on traditional handcrafted feature engineering. However, current deep learning models still face challenges in effectively capturing complex temporal dependencies in long-term time-series sensor data and addressing information redundancy, which affect model recognition accuracy and generalization ability. To address these issues, this paper proposes an innovative CNN-BiGRU–Transformer hybrid deep learning model aimed at improving the accuracy and robustness of human activity recognition. The proposed model integrates a multi-branch Convolutional Neural Network (CNN) to effectively extract multi-scale local spatial features, and combines a Bidirectional Gated Recurrent Unit (BiGRU) with a Transformer hybrid module for modeling temporal dependencies and extracting temporal features in long-term time-series data. Additionally, an attention mechanism is incorporated to dynamically allocate weights, suppress redundant information, and enhance key features, further improving recognition performance. To demonstrate the capability of the proposed model, evaluations are performed on three public datasets: WISDM, PAMAP2, and UCI-HAR. The model achieved recognition accuracies of 98.41%, 95.62%, and 96.74% on the three datasets, respectively, outperforming several state-of-the-art methods. These results confirm that the proposed approach effectively addresses feature extraction and redundancy challenges in long-term sensor time-series data and provides a robust solution for wearable sensor-based human activity recognition. Full article

Background: Artificial Intelligence (AI) techniques have been extensively deployed in sensor-based Human Activity Recognition (HAR) systems. Recent advances in deep learning, especially Convolutional Neural Networks (CNNs), have advanced HAR by enabling automatic feature extraction from raw sensor data. However, these models often struggle to capture multiscale patterns in human activity, limiting recognition accuracy. Additionally, traditional centralized learning approaches raise data privacy concerns, as personal sensor data must be transmitted to a central server, increasing the risk of privacy breaches. Methods: To address these challenges, this paper introduces Conv-ScaleNet, a CNN-based model designed for multiscale feature learning and compatibility with federated learning (FL) environments. Conv-ScaleNet integrates a Pyramid Pooling Module to extract both fine-grained and coarse-grained features and employs sequential Global Average Pooling layers to progressively capture abstract global representations from inertial sensor data. The model supports federated learning by training locally on user devices, sharing only model updates rather than raw data, thus preserving user privacy. Results: Experimental results demonstrate that the proposed Conv-ScaleNet achieves approximately 98% and 96% F1-scores on the WISDM and UCI-HAR datasets, respectively, confirming its competitiveness in FL environments for activity recognition. Conclusions: The proposed Conv-ScaleNet model addresses key limitations of existing HAR systems by combining multiscale feature learning with privacy-preserving training. Its strong performance, data protection capability, and adaptability to decentralized environments make it a robust and scalable solution for real-world HAR applications. Full article

Interpreting complex clinical time series is vital for patient safety and care, as it is both essential for supporting accurate clinical assessment and fundamental to building clinician trust and promoting effective clinical action. In complex time series analysis, decomposing a signal into meaningful underlying components is often a crucial means for achieving interpretability. This process is known as time series disentanglement. While deep learning models excel in predictive performance in this domain, their inherent complexity poses a major challenge to interpretability. Furthermore, existing time series disentanglement methods, including traditional trend or seasonality decomposition techniques, struggle to adequately separate clinically crucial specific components: static patient characteristics, condition trend, and acute events. Thus, a key technical challenge remains: developing an interpretable method capable of effectively disentangling these specific components in complex clinical time series. To address this challenge, we propose CoTD-VAE, a novel variational autoencoder framework for interpretable component disentanglement. CoTD-VAE incorporates temporal constraints tailored to the properties of static, trend, and event components, such as leveraging a Trend Smoothness Loss to capture gradual changes and an Event Sparsity Loss to identify potential acute events. These designs help the model effectively decompose time series into dedicated latent representations. We evaluate CoTD-VAE on critical care (MIMIC-IV) and human activity recognition (UCI HAR) datasets. Results demonstrate successful component disentanglement and promising performance enhancement in downstream tasks. Ablation studies further confirm the crucial role of our proposed temporal constraints. CoTD-VAE offers a promising interpretable framework for analyzing complex time series in critical applications like healthcare. Full article

Background: Activities of Daily Living (ADLs) are essential tasks performed at home and used in healthcare to monitor sedentary behavior, track rehabilitation therapy, and monitor chronic obstructive pulmonary disease. The Barthel Index, used by healthcare professionals, has limitations due to its subjectivity. Human activity recognition (HAR) is a more accurate method using Information and Communication Technologies (ICTs) to assess ADLs more accurately. This work aims to create a singular, adaptable, and heterogeneous ADL dataset that integrates information from various sources, ensuring a rich representation of different individuals and environments. Methods: A literature review was conducted in Scopus, the University of California Irvine (UCI) Machine Learning Repository, Google Dataset Search, and the University of Cauca Repository to obtain datasets related to ADLs. Inclusion criteria were defined, and a list of dataset characteristics was made to integrate multiple datasets. Twenty-nine datasets were identified, including data from various accelerometers, gyroscopes, inclinometers, and heart rate monitors. These datasets were classified and analyzed from the review. Tasks such as dataset selection, categorization, analysis, cleaning, normalization, and data integration were performed. Results: The resulting unified dataset contained 238,990 samples, 56 activities, and 52 columns. The integrated dataset features a wealth of information from diverse individuals and environments, improving its adaptability for various applications. Conclusions: In particular, it can be used in various data science projects related to ADL and HAR, and due to the integration of diverse data sources, it is potentially useful in addressing bias in and improving the generalizability of machine learning models. Full article

Continuous user authentication is critical to mobile device security, addressing vulnerabilities associated with traditional one-time authentication methods. This research proposes a hybrid deep learning framework that combines techniques from computer vision and sequence modeling, namely, ViT-inspired patch extraction, multi-head attention, and BiLSTM networks, to authenticate users continuously from smartphone sensor data. Unlike many existing approaches that directly apply these techniques for specific recognition tasks, our method reshapes raw motion signals into ViT-like patches to capture short-range patterns, employs multi-head attention to emphasize the most discriminative temporal segments, and then processes these enhanced embeddings through a bidirectional LSTM to integrate broader contextual information. This integrated pipeline effectively extracts local and global motion features specific to each user’s unique behavior, improving accuracy over conventional Transformer, Informer, CNN, and LSTM baselines. Experiments on the MotionSense and UCI HAR datasets show accuracies of 97.51% and 89.37%, respectively, indicating strong user-identification performance. Full article

Wearable sensors for human activity recognition (HAR) have gained significant attention across multiple domains, such as personal health monitoring and intelligent home systems. Despite notable advancements in deep learning for HAR, understanding the decision-making process of complex models remains challenging. This study introduces an advanced deep residual network integrated with a squeeze-and-excitation (SE) mechanism to improve recognition accuracy and model interpretability. The proposed model, ConvResBiGRU-SE, was tested using the UCI-HAR and WISDM datasets. It achieved remarkable accuracies of 99.18% and 98.78%, respectively, surpassing existing state-of-the-art methods. The SE mechanism enhanced the model’s ability to focus on essential features, while gradient-weighted class activation mapping (Grad-CAM) increased interpretability by highlighting essential sensory data influencing predictions. Additionally, ablation experiments validated the contribution of each component to the model’s overall performance. This research advances HAR technology by offering a more transparent and efficient recognition system. The enhanced transparency and predictive accuracy may increase user trust and facilitate smoother integration into real-world applications. Full article

Background: Artificial intelligence (AI)-based automated human activity recognition (HAR) is essential in enhancing assistive technologies for disabled individuals, focusing on fall detection, tracking rehabilitation progress, and analyzing personalized movement patterns. It also significantly manages and grows multiple industries, such as surveillance, sports, and diagnosis. Methods: This paper proposes a novel strategy using a three-stage feature ensemble combining deep learning (DL) and machine learning (ML) for accurate and automatic classification of activity recognition. We develop a unique activity detection approach in this study by enhancing the state-of-the-art convolutional neural network (CNN) and bi-directional long short-term memory (BiLSTM) models with selective ML classifiers and an attention mechanism. Thus, we developed an ensemble activity recognition model, namely “Attention-CNN-BiLSTM with selective ML”. Results: Out of the nine ML models and four DL models, the top performers are selected and combined in three stages for feature extraction. The effectiveness of this three-stage ensemble strategy is evaluated utilizing various performance metrics and through three distinct experiments. Utilizing the publicly available datasets (i.e., the UCI-HAR dataset and WISDM), our approach has shown superior predictive accuracy (98.75% and 99.58%, respectively). When compared with other methods, namely CNN, LSTM, CNN-BiLSTM, and Attention-CNN-BiLSTM, our approach surpasses them in terms of effectiveness, accuracy, and practicability. Conclusions: We hope that this comprehensive activity recognition system may be augmented with an advanced disability monitoring and diagnosis system to facilitate predictive assistance and personalized rehabilitation strategies. Full article

The proliferation of wearable sensors and mobile devices has fueled advancements in human activity recognition (HAR), with growing importance placed on both accuracy and privacy preservation. In this paper, the author proposes a federated learning framework for HAR, leveraging a hybrid Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) model to enhance feature extraction and classification in decentralized environments. Utilizing three public datasets—UCI-HAR, HARTH, and HAR7+—which contain diverse sensor data collected from free-living activities, the proposed system is designed to address the inherent privacy risks associated with centralized data processing by deploying Federated Averaging for local model training. To optimize recognition accuracy, the author introduces a dual-feature extraction mechanism, combining convolutional blocks for capturing local patterns and a hybrid LSTM-GRU structure to detect complex temporal dependencies. Furthermore, the author integrates an attention mechanism to focus on significant global relationships within the data. The proposed system is evaluated on the three public datasets—UCI-HAR, HARTH, and HAR7+—achieving superior performance compared to recent works in terms of F1-score and recognition accuracy. The results demonstrate that the proposed approach not only provides high classification accuracy but also ensures privacy preservation, making it a scalable and reliable solution for real-world HAR applications in decentralized and privacy-conscious environments. This work showcases the potential of federated learning in transforming human activity recognition, combining advanced feature extraction methodologies and privacy-respecting frameworks to deliver robust, real-time activity classification. Full article

In this study, we propose a novel framework for time-series representation learning that integrates a learnable masking-augmentation strategy into a contrastive learning framework. Time-series data pose challenges due to their temporal dependencies and feature-extraction complexities. To address these challenges, we introduce a masking-based reconstruction approach within a contrastive learning context, aiming to enhance the model’s ability to learn discriminative temporal features. Our method leverages self-supervised learning to effectively capture both global and local patterns by strategically masking segments of the time-series data and reconstructing them, which aids in revealing nuanced temporal dependencies. We utilize learnable masking as a dynamic augmentation technique, which enables the model to optimize contextual relationships in the data and extract meaningful representations that are both context-aware and robust. Extensive experiments were conducted on multiple time-series datasets, including SleepEDF-78, 20, UCI-HAR, achieving improvements of 2%, 2.55%, and 3.89% each and similar performance on Epilepsy in accuracy over baseline methods. Our results show significant performance gains compared to existing methods, highlighting the potential of our framework to advance the field of time-series analysis by improving the quality of learned representations and enhancing downstream task performance. Full article

Mobile and wearable devices have revolutionized the field of continuous user activity monitoring. However, analyzing the vast and intricate data captured by the sensors of these devices poses significant challenges. Deep neural networks have shown remarkable accuracy in Human Activity Recognition (HAR), but their application on mobile and wearable devices is constrained by limited computational resources. To address this limitation, we propose a novel method called Knowledge Distillation for Human Activity Recognition (KD-HAR) that leverages the knowledge distillation technique to compress deep neural network models for HAR using inertial sensor data. Our approach transfers the acquired knowledge from high-complexity teacher models (state-of-the-art models) to student models with reduced complexity. This compression strategy allows us to maintain performance while keeping computational costs low. To assess the compression capabilities of our approach, we evaluate it using two popular databases (UCI-HAR and WISDM) comprising inertial sensor data from smartphones. Our results demonstrate that our method achieves competitive accuracy, even at compression rates ranging from 18 to 42 times the number of parameters compared to the original teacher model. Full article

Human activity recognition (HAR) typically uses wearable sensors to identify and analyze the time-series data they collect, enabling recognition of specific actions. As such, HAR is increasingly applied in human–computer interaction, healthcare, and other fields, making accurate and efficient recognition of various human activities. In recent years, deep learning methods have been extensively applied in sensor-based HAR, yielding remarkable results. However, complex HAR research, which involves specific human behaviors in varied contexts, still faces several challenges. To solve these problems, we propose a multi-head neural network based on the attention mechanism. This framework contains three convolutional heads, with each head designed using one-dimensional CNN to extract features from sensory data. The model uses a channel attention module (squeeze–excitation module) to enhance the representational capabilities of convolutional neural networks. We conducted experiments on two publicly available benchmark datasets, UCI-HAR and WISDM, to evaluate our model. The results were satisfactory, with overall recognition accuracies of 96.72% and 97.73% on their respective datasets. The experimental results demonstrate the effectiveness of the network structure for the HAR, which ensures a higher level of accuracy. Full article

Human activity recognition (HAR) is a crucial task in various applications, including healthcare, fitness, and the military. Deep learning models have revolutionized HAR, however, their computational complexity, particularly those involving BiLSTMs, poses significant challenges for deployment on resource-constrained devices like smartphones. While BiLSTMs effectively capture long-term dependencies by processing inputs bidirectionally, their high parameter count and computational demands hinder practical applications in real-time HAR. This study investigates the approximation of the computationally intensive BiLSTM component in a HAR model by using a combination of alternative model components and data flipping augmentation. The proposed modifications to an existing hybrid model architecture replace the BiLSTM with standard and residual LSTM, along with convolutional networks, supplemented by data flipping augmentation to replicate the context awareness typically provided by BiLSTM networks. The results demonstrate that the residual LSTM (ResLSTM) model achieves superior performance while maintaining a lower computational complexity compared to the traditional BiLSTM model. Specifically, on the UCI-HAR dataset, the ResLSTM model attains an accuracy of 96.34% with 576,702 parameters, outperforming the BiLSTM model’s accuracy of 95.22% with 849,534 parameters. On the WISDM dataset, the ResLSTM achieves an accuracy of 97.20% with 192,238 parameters, compared to the BiLSTM’s 97.23% accuracy with 283,182 parameters, demonstrating a more efficient architecture with minimal performance trade-off. For the KU-HAR dataset, the ResLSTM model achieves an accuracy of 97.05% with 386,038 parameters, showing comparable performance to the BiLSTM model’s 98.63% accuracy with 569,462 parameters, but with significantly fewer parameters. Full article

Sensor-based human activity recognition has been widely used in various fields; however, there are still challenges involving recognition of daily complex human activities using sensors. In order to solve the problem of timeliness and homogeneity of recognition functions in human activity recognition models, we propose a human activity recognition model called ’BSTCA-HAR’ based on a long short-term memory (LSTM) network. The approach proposed in this paper combines an attention mechanism and a temporal convolutional network (TCN). The learning and prediction units in the model can efficiently learn important action data while capturing long time-dependent information as well as features at different time scales. Our series of experiments on three public datasets (WISDM, UCI-HAR, and ISLD) with different data features confirm the feasibility of the proposed method. This method excels in dynamically capturing action features while maintaining a low number of parameters and achieving a remarkable average accuracy of $93\%$, proving that the model has good recognition performance. Full article

Human activity recognition (HAR) based on wearable sensors has emerged as a low-cost key-enabling technology for applications such as human–computer interaction and healthcare. In wearable sensor-based HAR, deep learning is desired for extracting human active features. Due to the spatiotemporal dynamic of human activity, a special deep learning network for recognizing the temporal continuous activities of humans is required to improve the recognition accuracy for supporting advanced HAR applications. To this end, a residual multifeature fusion shrinkage network (RMFSN) is proposed. The RMFSN is an improved residual network which consists of a multi-branch framework, a channel attention shrinkage block (CASB), and a classifier network. The special multi-branch framework utilizes a 1D-CNN, a lightweight temporal attention mechanism, and a multi-scale feature extraction method to capture diverse activity features via multiple branches. The CASB is proposed to automatically select key features from the diverse features for each activity, and the classifier network outputs the final recognition results. Experimental results have shown that the accuracy of the proposed RMFSN for the public datasets UCI-HAR, WISDM, and OPPORTUNITY are 98.13%, 98.35%, and 93.89%, respectively. In comparison with existing advanced methods, the proposed RMFSN could achieve higher accuracy while requiring fewer model parameters. Full article

Learning underlying patterns from sensory data is crucial in the Human Activity Recognition (HAR) task to avoid poor generalization when coping with unseen data. A key solution to such an issue is representation learning, which becomes essential when input signals contain activities with similar patterns or when patterns generated by different subjects for the same activity vary. To address these issues, we seek a solution to increase generalization by learning the underlying factors of each sensor signal. We develop a novel multi-channel asymmetric auto-encoder to recreate input signals precisely and extract indicative unsupervised futures. Further, we investigate the role of various activation functions in signal reconstruction to ensure the model preserves the patterns of each activity in the output. Our main contribution is that we propose a multi-task learning model to enhance representation learning through shared layers between signal reconstruction and the HAR task to improve the robustness of the model in coping with users not included in the training phase. The proposed model learns shared features between different tasks that are indeed the underlying factors of each input signal. We validate our multi-task learning model using several publicly available HAR datasets, UCI-HAR, MHealth, PAMAP2, and USC-HAD, and an in-house alpine skiing dataset collected in the wild, where our model achieved 99%, 99%, 95%, 88%, and 92% accuracy. Our proposed method shows consistent performance and good generalization on all the datasets compared to the state of the art. Full article