A Flexible Approach for Human Activity Recognition Using Artificial Hydrocarbon Networks

Recognition of sports and daily activities through deep learning and convolutional block attention

Portable devices like accelerometers and physiological trackers capture movement and biometric data relevant to sports. This study uses data from wearable sensors to investigate deep learning techniques for recognizing human behaviors associated with sports and fitness. The proposed CNN-BiGRU-CBAM model, a unique hybrid architecture, combines convolutional neural networks (CNNs), bidirectional gated recurrent unit networks (BiGRUs), and convolutional block attention modules (CBAMs) for accurate activity recognition. CNN layers extract spatial patterns, BiGRU captures temporal context, and CBAM focuses on informative BiGRU features, enabling precise activity pattern identification. The novelty lies in seamlessly integrating these components to learn spatial and temporal relationships, prioritizing significant features for activity detection. The model and baseline deep learning models were trained on the UCI-DSA dataset, evaluating with 5-fold cross-validation, including multi-class classification accuracy, precision, recall, and F1-score. The CNN-BiGRU-CBAM model outperformed baseline models like CNN, LSTM, BiLSTM, GRU, and BiGRU, achieving state-of-the-art results with 99.10% accuracy and F1-score across all activity classes. This breakthrough enables accurate identification of sports and everyday activities using simplified wearables and advanced deep learning techniques, facilitating athlete monitoring, technique feedback, and injury risk detection. The proposed model's design and thorough evaluation significantly advance human activity recognition for sports and fitness.

Intelligent Diagnosis Based on Double-Optimized Artificial Hydrocarbon Networks for Mechanical Faults of In-Wheel Motor

To avoid the potential safety hazards of electric vehicles caused by the mechanical fault deterioration of the in-wheel motor (IWM), this paper proposes an intelligent diagnosis based on double-optimized artificial hydrocarbon networks (AHNs) to identify the mechanical faults of IWM, which employs a K-means clustering and AdaBoost algorithm to solve the lower accuracy and poorer stability of traditional AHNs. Firstly, K-means clustering is used to improve the interval updating method of any adjacent AHNs molecules, and then simplify the complexity of the AHNs model. Secondly, the AdaBoost algorithm is utilized to adaptively distribute the weights for multiple weak models, then reconstitute the network structure of the AHNs. Finally, double-optimized AHNs are used to build an intelligent diagnosis system, where two cases of bearing datasets from Paderborn University and a self-made IWM test stand are processed to validate the better performance of the proposed method, especially in multiple rotating speeds and the load conditions of the IWM. The double-optimized AHNs provide a higher accuracy for identifying the mechanical faults of the IWM than the traditional AHNs, K-means-based AHNs (K-AHNs), support vector machine (SVM), and particle swarm optimization-based SVM (PSO-SVM).

Intelligent diagnosis of mechanical faults of in-wheel motor based on improved artificial hydrocarbon networks

For the driving safety of electric vehicle (EV), intelligent diagnosis based on artificial hydrocarbon networks (AHNs) is proposed to detect mechanical faults of in-wheel motor (IWM) which is a promising force pattern of EV. AHNs, a novel mathematical model of supervised learning algorithm, can encapsulate or inherit or mix any information, then are adapted to deal with serious external interference and the variable operating conditions. Based on the basic AHNs, complex error function is proposed to optimize more information of classification targets, and distance error ratio is defined to evaluate the performance. Then, the improved AHNs is employed to build two intelligent diagnosis systems namely one-stop diagnosis and sequential diagnosis, which select the same and different symptom parameters as the object of a follow-on process, respectively. The effectiveness of the proposed methods is validated by two case studies of Case Western Reserve University dataset and mechanical faults data from IWM's test bench.

Understanding LSTM Network Behaviour of IMU-Based Locomotion Mode Recognition for Applications in Prostheses and Wearables

Human Locomotion Mode Recognition (LMR) has the potential to be used as a control mechanism for lower-limb active prostheses. Active prostheses can assist and restore a more natural gait for amputees, but as a medical device it must minimize user risks, such as falls and trips. As such, any control system must have high accuracy and robustness, with a detailed understanding of its internal operation. Long Short-Term Memory (LSTM) machine-learning networks can perform LMR with high accuracy levels. However, the internal behavior during classification is unknown, and they struggle to generalize when presented with novel users. The target problem addressed in this paper is understanding the LSTM classification behavior for LMR. A dataset of six locomotive activities (walking, stopped, stairs and ramps) from 22 non-amputee subjects is collected, capturing both steady-state and transitions between activities in natural environments. Non-amputees are used as a substitute for amputees to provide a larger dataset. The dataset is used to analyze the internal behavior of a reduced complexity LSTM network. This analysis identifies that the model primarily classifies activity type based on data around early stance. Evaluation of generalization for unseen subjects reveals low sensitivity to hyper-parameters and over-fitting to individuals’ gait traits. Investigating the differences between individual subjects showed that gait variations between users primarily occur in early stance, potentially explaining the poor generalization. Adjustment of hyper-parameters alone could not solve this, demonstrating the need for individual personalization of models. The main achievements of the paper are (i) the better understanding of LSTM for LMR, (ii) demonstration of its low sensitivity to learning hyper-parameters when evaluating novel user generalization, and (iii) demonstration of the need for personalization of ML models to achieve acceptable accuracy.

Sport-Related Human Activity Detection and Recognition Using a Smartwatch

As an active research field, sport-related activity monitoring plays an important role in people’s lives and health. This is often viewed as a human activity recognition task in which a fixed-length sliding window is used to segment long-term activity signals. However, activities with complex motion states and non-periodicity can be better monitored if the monitoring algorithm is able to accurately detect the duration of meaningful motion states. However, this ability is lacking in the sliding window approach. In this study, we focused on two types of activities for sport-related activity monitoring, which we regard as a human activity detection and recognition task. For non-periodic activities, we propose an interval-based detection and recognition method. The proposed approach can accurately determine the duration of each target motion state by generating candidate intervals. For weak periodic activities, we propose a classification-based periodic matching method that uses periodic matching to segment the motion sate. Experimental results show that the proposed methods performed better than the sliding window method.

Multi-Sensor Fusion for Activity Recognition-A Survey

In Ambient Intelligence (AmI), the activity a user is engaged in is an essential part of the context, so its recognition is of paramount importance for applications in areas like sports, medicine, personal safety, and so forth. The concurrent use of multiple sensors for recognition of human activities in AmI is a good practice because the information missed by one sensor can sometimes be provided by the others and many works have shown an accuracy improvement compared to single sensors. However, there are many different ways of integrating the information of each sensor and almost every author reporting sensor fusion for activity recognition uses a different variant or combination of fusion methods, so the need for clear guidelines and generalizations in sensor data integration seems evident. In this survey we review, following a classification, the many fusion methods for information acquired from sensors that have been proposed in the literature for activity recognition; we examine their relative merits, either as they are reported and sometimes even replicated and a comparison of these methods is made, as well as an assessment of the trends in the area.

Virtual Sensors for Optimal Integration of Human Activity Data

Sensors are becoming more and more ubiquitous as their price and availability continue to improve, and as they are the source of information for many important tasks. However, the use of sensors has to deal with noise and failures. The lack of reliability in the sensors has led to many forms of redundancy, but simple solutions are not always the best, and the precise way in which several sensors are combined has a big impact on the overall result. In this paper, we discuss how to deal with the combination of information coming from different sensors, acting thus as “virtual sensors”, in the context of human activity recognition, in a systematic way, aiming for optimality. To achieve this goal, we construct meta-datasets containing the “signatures” of individual datasets, and apply machine-learning methods in order to distinguish when each possible combination method could be actually the best. We present specific results based on experimentation, supporting our claims of optimality.

Pulsar Detection for Wavelets SODA and Regularized Fuzzy Neural Networks Based on Andneuron and Robust Activation Function

The use of intelligent models may be slow because of the number of samples involved in the problem. The identification of pulsars (stars that emit Earth-catchable signals) involves collecting thousands of signals by professionals of astronomy and their identification may be hampered by the nature of the problem, which requires many dimensions and samples to be analyzed. This paper proposes the use of hybrid models based on concepts of regularized fuzzy neural networks that use the representativeness of input data to define the groupings that make up the neurons of the initial layers of the model. The andneurons are used to aggregate the neurons of the first layer and can create fuzzy rules. The training uses fast extreme learning machine concepts to generate the weights of neurons that use robust activation functions to perform pattern classification. To solve large-scale problems involving the nature of pulsar detection problems, the model proposes a fast and highly accurate approach to address complex issues. In the execution of the tests with the proposed model, experiments were conducted explanation in two databases of pulsars, and the results prove the viability of the fast and interpretable approach in identifying such involved stars.

An Intelligent Failure Detection on a Wireless Sensor Network for Indoor Climate Conditions

Wireless sensor networks (WSN) involve large number of sensor nodes distributed at diverse locations. The collected data are prone to be inaccurate and faulty due to internal or external influences, such as, environmental interference or sensor aging. Intelligent failure detection is necessary for the effective functioning of the sensor network. In this paper, we propose a supervised learning method that is named artificial hydrocarbon networks (AHN), to predict temperature in a remote location and detect failures in sensors. It allows predicting the temperature and detecting failure in sensor node of remote locations using information from a web service comparing it with field temperature sensors. For experimentation, we implemented a small WSN to test our sensor in order to measure failure detection, identification and accommodation proposal. In our experiments, 94.18% of the testing data were recovered and accommodated allowing of validation our proposed approach that is based on AHN, which detects, identify and accommodate sensor failures accurately.

Comparison of Different Sets of Features for Human Activity Recognition by Wearable Sensors

Human Activity Recognition (HAR) refers to an emerging area of interest for medical, military, and security applications. However, the identification of the features to be used for activity classification and recognition is still an open point. The aim of this study was to compare two different feature sets for HAR. Particularly, we compared a set including time, frequency, and time-frequency domain features widely used in literature (FeatSet_A) with a set of time-domain features derived by considering the physical meaning of the acquired signals (FeatSet_B). The comparison of the two sets were based on the performances obtained using four machine learning classifiers. Sixty-one healthy subjects were asked to perform seven different daily activities wearing a MIMU-based device. Each signal was segmented using a 5-s window and for each window, 222 and 221 variables were extracted for the FeatSet_A and FeatSet_B respectively. Each set was reduced using a Genetic Algorithm (GA) simultaneously performing feature selection and classifier optimization. Our results showed that Support Vector Machine achieved the highest performances using both sets (97.1% and 96.7% for FeatSet_A and FeatSet_B respectively). However, FeatSet_B allows to better understand alterations of the biomechanical behavior in more complex situations, such as when applied to pathological subjects.

Impact of Sliding Window Length in Indoor Human Motion Modes and Pose Pattern Recognition Based on Smartphone Sensors

Human activity recognition (HAR) is essential for understanding people’s habits and behaviors, providing an important data source for precise marketing and research in psychology and sociology. Different approaches have been proposed and applied to HAR. Data segmentation using a sliding window is a basic step during the HAR procedure, wherein the window length directly affects recognition performance. However, the window length is generally randomly selected without systematic study. In this study, we examined the impact of window length on smartphone sensor-based human motion and pose pattern recognition. With data collected from smartphone sensors, we tested a range of window lengths on five popular machine-learning methods: decision tree, support vector machine, K-nearest neighbor, Gaussian naïve Bayesian, and adaptive boosting. From the results, we provide recommendations for choosing the appropriate window length. Results corroborate that the influence of window length on the recognition of motion modes is significant but largely limited to pose pattern recognition. For motion mode recognition, a window length between 2.5–3.5 s can provide an optimal tradeoff between recognition performance and speed. Adaptive boosting outperformed the other methods. For pose pattern recognition, 0.5 s was enough to obtain a satisfactory result. In addition, all of the tested methods performed well.

A Comparison Study of Classifier Algorithms for Cross-Person Physical Activity Recognition

Physical activity is widely known to be one of the key elements of a healthy life. The many benefits of physical activity described in the medical literature include weight loss and reductions in the risk factors for chronic diseases. With the recent advances in wearable devices, such as smartwatches or physical activity wristbands, motion tracking sensors are becoming pervasive, which has led to an impressive growth in the amount of physical activity data available and an increasing interest in recognizing which specific activity a user is performing. Moreover, big data and machine learning are now cross-fertilizing each other in an approach called “deep learning”, which consists of massive artificial neural networks able to detect complicated patterns from enormous amounts of input data to learn classification models. This work compares various state-of-the-art classification techniques for automatic cross-person activity recognition under different scenarios that vary widely in how much information is available for analysis. We have incorporated deep learning by using Google’s TensorFlow framework. The data used in this study were acquired from PAMAP2 (Physical Activity Monitoring in the Ageing Population), a publicly available dataset containing physical activity data. To perform cross-person prediction, we used the leave-one-subject-out (LOSO) cross-validation technique. When working with large training sets, the best classifiers obtain very high average accuracies (e.g., 96% using extra randomized trees). However, when the data volume is drastically reduced (where available data are only 0.001% of the continuous data), deep neural networks performed the best, achieving 60% in overall prediction accuracy. We found that even when working with only approximately 22.67% of the full dataset, we can statistically obtain the same results as when working with the full dataset. This finding enables the design of more energy-efficient devices and facilitates cold starts and big data processing of physical activity records.