Human activity recognition (HAR) based on wearable devices has become an active research direction in the field of ubiquitous computing, and has a wide range of Internet of Things (IoT) applications. Unfortunately, it is challenging to obtain large amounts of labeled sensing data, and manual annotation is time-consuming and labor-intensive, making it impossible for the extensive deployment of HAR systems. Consequently, self-supervised learning has emerged to address this challenge by training on unlabeled data. However, traditional contrastive learning fails to simulate more sample diversity problems caused by environmental heterogeneity and sensor heterogeneity. In this paper, we propose a multi-positive sample quantum contrastive learning (MPSQCL) framework. By increasing the positive samples for contrastive learning and leveraging the advantages of quantum machine learning (QML) techniques, the richer features of input samples are extracted to improve the robustness and generalization of the model. Moreover, we design a new contrastive loss function to adapt to multiple positive sample contrastive learning scenarios. Finally, we validate the effectiveness of the proposed framework on several publicly available HAR datasets.
Human activity recognition (HAR) based on wearable devices has become an active research direction in the field of ubiquitous computing, and has a wide range of Internet of Things (IoT) applications. Unfortunately, it is challenging to obtain large amounts of labeled sensing data, and manual annotation is time-consuming and labor-intensive, making it impossible for the extensive deployment of HAR systems. Consequently, self-supervised learning has emerged to address this challenge by training on unlabeled data. However, traditional contrastive learning fails to simulate more sample diversity problems caused by environmental heterogeneity and sensor heterogeneity. In this paper, we propose a multi-positive sample quantum contrastive learning (MPSQCL) framework. By increasing the positive samples for contrastive learning and leveraging the advantages of quantum machine learning (QML) techniques, the richer features of input samples are extracted to improve the robustness and generalization of the model. Moreover, we design a new contrastive loss function to adapt to multiple positive sample contrastive learning scenarios. Finally, we validate the effectiveness of the proposed framework on several publicly available HAR datasets.
Quantum machine learning, as an extension of classical machine learning that harnesses quantum mechanics, facilitates effiient learning from data encoded in quantum states. Training a quantum neural network typically demands a substantial labeled training set for supervised learning. Human annotators, often experts, provide labels for samples through additional experiments, adding to the training cost. To mitigate this expense, there is a quest for methods that maintain model performance over fully labeled datasets while requiring fewer labeled samples in practice, thereby extending few-shot learning to the quantum realm. Quantum active learning estimates the uncertainty of quantum data to select the most informative samples from a pool for labeling. Consequently, a QML model is supposed to accumulate maximal knowledge as the training set comprises labeled samples selected via sampling strategies. Notably, the QML models trained within the QAL framework are not restricted to specific types, enabling performance enhancement from the model architecture's perspective towards few-shot learning. Recognizing symmetry as a fundamental concept in physics ubiquitous across various domains, we leverage the symmetry inherent in quantum states induced by the embedding of classical data for model design. We employ an equivariant QNN capable of generalizing from fewer data with geometric priors. We benchmark the performance of QAL on two classification problems, observing both positive and negative results. QAL effectively trains the model, achieving performance comparable to that on fully labeled datasets by labeling less than 7\% of the samples in the pool with unbiased sampling behavior. Furthermore, we elucidate the negative result of QAL being overtaken by random sampling baseline through miscellaneous numerical experiments. (character count limit, see the main text)
The rapid development and integration of wearable technologies or Internet-of-Things (IoT) into healthcare technologies with computational methods holds transformative potential for future healthcare systems, especially for personalized health monitoring and early disease detection. In parallel, recent advancements in quantum computing have o↵ered significant interest in quantum machine learning (QML), an intersection of quantum computing and machine learning, which provides a novel hybrid computational paradigm capable of addressing complex healthcare challenges that remain unsolvable with classical machine learning techniques. This thesis explores the application of quantum-enhanced recurrent neural network architectures, specifically Quantum Long Short-Term Memory (QLSTM) and Quantum-based Gated Recurrent Unit (QGRU), for predictive modeling and activity classification in wearable IoT systems, focusing particularly on their applicability and efficacy to elderly populations. One of the key challenges in modern healthcare intelligence systems is the simple but accurate estimation mechanism of Physical Activity Energy Expenditure (PAEE), which is an essential factor for assessing physical health status, particularly for healthy aging. In order to address this challenge, the first work proposes a model called an enhanced QLSTM with linear layer (eQLSTML) that integrates Variational Quantum Circuits (VQCs) into classical LSTM architectures. This integration notably explores the advantages of quantum computing, including properties such as entanglement and superposition, to enhance the model’s ability to capture complex temporal dependencies and subtle variations in human activity patterns. The modelling experiments were evaluated using the publicly available GOTOV Human Physical Activity dataset, which includes accelerometer data collected from older adults engaged in various daily activities. In addition to modelling architecture analysis, this thesis provides comprehensive analyses concerning the scalability, feasibility, and computational complexity of the proposed quantum-enhanced models compared to the classical backbone. The overall results demonstrate that the proposed quantum-enhanced approach significantly outperforms traditional classical machine learning algorithms, which demonstrates the potential ability for accurate predictions and robustness to noisy data like accelerometer data. In addition, accurate classification of activities of daily living (ADLs) is another critical aspect for early disease detection and e↵ective healthcare intervention. Within the context of this thesis, the second work introduces a novel hybrid model called QGRU-Multiclass Classifier (QGRU-MC). The data preprocessing involves creating a statistical feature extraction and applying oversampling techniques – SMOTE-ENN (Synthetic Minority Oversampling Technique with Edited Nearest Neighbors) for the publicly available dataset called “Dataset for ADL recognition with Wrist-worn Accelerometer”. The preliminary findings suggest that our model has good potential for healthcare applications, particularly in advancing future intelligent systems focused on daily activity monitoring. Specifically, QGRU-MC achieved significant improvements in classification accuracy, sensitivity, and specificity across multiple categories of daily activities, highlighting its potential for practical deployment in intelligent wearable systems aimed at monitoring and supporting healthy aging. Furthermore, this work also discusses the future research directions and practical considerations for transforming from the simulators to modern real quantum computers in the noisy intermediate-scale quantum (NISQ) era for a more comprehensive view on exploiting potential impacts on public health policies and personalized healthcare systems. Overall, simulator-based results indicate promising potential for integrating QML methods into wearable healthcare applications such as predictive analytics, personalized treatments, and health disease intervention to improve healthcare outcomes. Specifically, on PAEE regression, the proposed eQLSTML improved R² by 13.2% and reduced RMSE and MAE by 14.8% and 19.1%, respectively, relative to the classical LSTM; on ADL, QGRU-MC demonstrates good macro-averaged metrics across all eight classes (see Chs. 3–4). This research contributes significantly to the growing body of literature supporting quantum computing technologies in healthcare applications, making QML an incredible tool for addressing complex healthrelated computational challenges in the era of big and complex data in medical fields and smart healthcare technologies, especially in e↵ectively solving data from multiple sources, wearable accelerometry, and multi-class ADL recognition with complicated tasks.
Quantum machine learning (QML) has attracted growing interest with the rapid parallel advances in large-scale classical machine learning and quantum technologies. Similar to classical machine learning, QML models also face challenges arising from the scarcity of labeled data, particularly as their scale and complexity increase. Here, we introduce self-supervised pretraining of quantum representations that reduces reliance on labeled data by learning invariances from unlabeled examples. We implement this paradigm on a programmable trapped-ion quantum computer, encoding images as quantum states. In situ contrastive pretraining on hardware yields a representation that, when fine-tuned, classifies image families with higher mean test accuracy and lower run-to-run variability than models trained from random initialization. Performance improvement is especially significant in regimes with limited labeled training data. We show that the learned invariances generalize beyond the pretraining image samples. Unlike prior work, our pipeline derives similarity from measured quantum overlaps and executes all training and classification stages on hardware. These results establish a label-efficient route to quantum representation learning, with direct relevance to quantum-native datasets and a clear path to larger classical inputs.
Human Activity Recognition (HAR) plays an important role in smart environments and Internet of Things (IoT) applications, where accurate real-time monitoring can enhance automation and interaction. While Classical Convolutional Neural Networks (CNNs) have been widely used for such tasks, they face limitations in scalability and feature extraction, especially when dealing with large datasets. To address these challenges, we propose a novel application of Quantum Convolutional Neural Networks (QCNNs) specifically tailored for processing CSI data in HAR. Our method aims to combine the power of quantum computing with classical convolutional methods to extract complex patterns from CSI signals, thereby improving accuracy and maintaining computational efficiency. The challenge of high-dimensional CSI data is addressed by hybridizing classical and quantum layers in the architecture, which enables better feature extraction and learning capabilities. Results show that our QCNN achieves superior performance in comparison to traditional CNNs.
Wi-Fi-based human activity recognition (HAR) has become an important part of integrated sensing and communications, paving the way for a range of context-aware services. However, most existing Wi-Fi-based HAR systems rely on deep learning (DL) models that are computationally and memory intensive in both training and inference, which poses significant challenges for real-world deployment. Conventional training requires simultaneous updates of millions of parameters, leading to prohibitive memory consumption. In this paper, we propose a novel quantum-assisted memory-efficient training framework (Q-MET) designed to improve efficiency in both training and inference. Q-MET utilizes a hybrid quantum classical neural network to indirectly generate parameters for HAR models, significantly reducing the trainable parameter count compared to direct optimization. To further support the deployment on resource-constrained devices, we integrate structured pruning during the training phase. Experimental results demonstrate that Q-MET achieves a 90% to 95% reduction in trainable parameters compared with conventional backpropagation-based DL training while maintaining or even exceeding classical classification accuracy. Additionally, Q-MET supports lightweight inference through structured pruning, achieving 75% to 85% model sparsity with less than 2% loss in classification accuracy. To the best of our knowledge, this work represents the first quantum-assisted approach to simultaneously tackle memory inefficiencies in both the training and inference stages of HAR systems.
In the evolving landscape of data privacy regulations, the challenge of providing extensive data for robust deep classification models arises. The accuracy of these models relies on the amount of training data, due to the multitude of parameters that require tuning. Unfortunately, obtaining such ample data proves challenging, particularly in domains like medical applications, where there is a pressing need for robust models for early disease detection but a shortage of labeled data. Nevertheless, the classical supervised contrastive learning models, have shown the potential to address this challenge up to a certain limit, by utilizing deep encoder models. However, recent advancements in quantum machine learning enable the extraction of meaningful representations from extremely limited and simple data. Thus, replacing classical counterparts in classical or hybrid quantum-classical supervised contrastive models enhances feature learning capability with minimal data. Therefore, this work proposes the Q-SupCon model, a fully quantum-powered supervised contrastive learning model comprising a quantum data augmentation circuit, quantum encoder, quantum projection head, and quantum variational classifier, enabling efficient image classification with minimal labeled data. Furthermore, the novel model attains 80%, 60%, and 80% test accuracy on MNIST, KMNIST, and FMNIST datasets, marking a significant advancement in addressing the data scarcity challenge.