Abstract:The rapid growth of wearable sensor technologies holds substantial promise for the field of personalized and context-aware Human Activity Recognition. Given the inherently decentralized nature of data sources within this domain, the utilization of multi-agent systems with their inherent decentralization capabilities presents an opportunity to facilitate the development of scalable, adaptable, and privacy-conscious methodologies. This paper introduces a collaborative distributed learning approach rooted in multi-agent principles, wherein individual users of sensor-equipped devices function as agents within a distributed network, collectively contributing to the comprehensive process of learning and classifying human activities. In this proposed methodology, not only is the privacy of activity monitoring data upheld for each individual, eliminating the need for an external server to oversee the learning process, but the system also exhibits the potential to surmount the limitations of conventional centralized models and adapt to the unique attributes of each user. The proposed approach has been empirically tested on two publicly accessible human activity recognition datasets, specifically PAMAP2 and HARTH, across varying settings. The provided empirical results conclusively highlight the efficacy of inter-individual collaborative learning when contrasted with centralized configurations, both in terms of local and global generalization.

What problem does this paper attempt to address?

The paper primarily aims to address key issues encountered in the field of Human Activity Recognition (HAR), particularly by enhancing privacy protection, adaptability, and scalability through a distributed, multi-agent system approach. Specifically, the research proposes an agent-based collaborative learning method to overcome the limitations of traditional centralized methods in terms of privacy protection, adapting to new individuals, and handling dynamic scenarios. ### Research Questions The core objectives of the paper are: 1. **Privacy Protection**: Enhancing user privacy by performing distributed learning on devices, thus avoiding the need to send raw data to a central server. 2. **Adaptability and Scalability**: Leveraging the characteristics of multi-agent systems to improve the model's adaptability to different users and scenarios, and to enhance the overall scalability of the system. 3. **Cross-Individual Learning**: Allowing data from different individuals to collaborate in order to improve the accuracy of activity recognition, especially in the face of non-uniform data distribution. ### Solution The paper proposes a novel distributed collaborative learning framework, where each individual equipped with sensors is considered an agent in the network. These agents independently train their own machine learning models and collaborate by exchanging model update information. This approach does not require a centralized server to oversee the entire learning process but relies on direct communication between agents. ### Main Contributions - **Fully Decentralized**: The method does not rely on a central server; all agents participate in model updates and collaboration. - **Flexibility and Scalability**: The method can adapt to different individuals and various types of sensor data while maintaining good privacy protection. - **Empirical Validation**: The paper conducts experiments on two public human activity recognition datasets (PAMAP2 and HARTH) to validate the effectiveness of the proposed distributed method in both local and global generalization capabilities. ### Conclusion By introducing an agent-based collaborative learning method, the study successfully addresses some key challenges in the field of human activity recognition, particularly in terms of privacy protection, adaptability, and scalability. Moreover, experimental results demonstrate that the method not only improves model performance but also shows strong potential in dealing with imbalanced data distribution.

Distributed Agent-Based Collaborative Learning in Cross-Individual Wearable Sensor-Based Human Activity Recognition

Wearable Sensor Based Multimodal Human Activity Recognition Exploiting the Diversity of Classifier Ensemble.

Continual learning in sensor-based human activity recognition: An empirical benchmark analysis

Towards Deep Clustering of Human Activities from Wearables

Collaborative Activity Recognition with Heterogeneous Activity Sets and Privacy Preferences

SLearn: Shared Learning Human Activity Labels Across Multiple Datasets.

Context-aware mutual learning for semi-supervised human activity recognition using wearable sensors

Online Continual Learning for Human Activity Recognition

Multimodal Human Activity Recognition From Wearable Inertial Sensors Using Machine Learning

Recognizing Multi-User Activities Using Wearable Sensors in a Smart Home

A Profile Similarity-Based Personalized Federated Learning Method for Wearable Sensor-Based Human Activity Recognition

Decoding Human Activities: Analyzing Wearable Accelerometer and Gyroscope Data for Activity Recognition

Wearable Multi-Sensor Data Fusion Approach for Human Activity Recognition Using Machine Learning Algorithms

Privacy-Preserving Learning of Human Activity Predictors in Smart Environments

Personalized Human Activity Recognition Based on Integrated Wearable Sensor and Transfer Learning.

Human Activity Recognition Using Self-Supervised Representations of Wearable Data

Activation‐based recurrent learning method for wearable sensor data processing in human activity recognition

A holistic multi-source transfer learning approach using wearable sensors for personalized daily activity recognition

Federated recognition mechanism based on enhanced temporal-spatial learning using mobile edge sensors for firefighters

TASKED: Transformer-based Adversarial learning for human activity recognition using wearable sensors via Self-KnowledgE Distillation

Privacy-Preserving Collaborative Computation for Human Activity Recognition