The 8th EAI International Conference on Mobile Computing, Applications and Services

Research Article

Hybridizing Personal and Impersonal Machine Learning Models for Activity Recognition on Mobile Devices

  • @INPROCEEDINGS{10.4108/eai.30-11-2016.2267108,
        author={Tong Yu and Yong Zhuang and Ole Mengshoel and Osman Yagan},
        title={Hybridizing Personal and Impersonal Machine Learning Models for Activity Recognition on Mobile Devices},
        proceedings={The 8th EAI International Conference on Mobile Computing, Applications and Services},
        publisher={ACM},
        proceedings_a={MOBICASE},
        year={2016},
        month={12},
        keywords={activity recognition incremental learning adaptive learning rate cost-sensitive privacy logistic regression},
        doi={10.4108/eai.30-11-2016.2267108}
    }
    
  • Tong Yu
    Yong Zhuang
    Ole Mengshoel
    Osman Yagan
    Year: 2016
    Hybridizing Personal and Impersonal Machine Learning Models for Activity Recognition on Mobile Devices
    MOBICASE
    ACM
    DOI: 10.4108/eai.30-11-2016.2267108
Tong Yu1,*, Yong Zhuang1, Ole Mengshoel1, Osman Yagan1
  • 1: Carnegie Mellon University
*Contact email: tong.yu@sv.cmu.edu

Abstract

Recognition of human activities, using smart phones and wearable devices, has attracted much attention recently. The machine learning (ML) approach to human activity recognition can broadly be classified into two categories: training an ML model on (i) an impersonal dataset or (ii) a personal dataset. Previous research shows that models learned from personal datasets can provide better activity recognition accuracy compared to models trained on impersonal datasets. In this paper, we develop a hybrid incremental (HI) method with logistic regression models. This method uses incremental learning of logistic regression, to combine the advantages of both impersonal and personal approaches. We investigate two essential issues in this method, which are the selection of the learning rate schedule and the class imbalance problem. Our experiments show that the model learned using our HI method give better accuracy than the model learned from personal or impersonal data only. Besides, the techniques of adaptive learning rate and cost-sensitive learning give faster updates and more robust ML models in incremental learning. Our method also has potential benefits in the area of privacy preservation.