A User-Adaptive Algorithm for Activity Recognition Based on K-Means Clustering, Local Outlier Factor, and Multivariate Gaussian Distribution

Abstract

Mobile activity recognition is significant to the development of human-centric pervasive applications including elderly care, personalized recommendations, etc. Nevertheless, the distribution of inertial sensor data can be influenced to a great extent by varying users. This means that the performance of an activity recognition classifier trained by one user’s dataset will degenerate when transferred to others. In this study, we focus on building a personalized classifier to detect four categories of human activities: light intensity activity, moderate intensity activity, vigorous intensity activity, and fall. In order to solve the problem caused by different distributions of inertial sensor signals, a user-adaptive algorithm based on K-Means clustering, local outlier factor (LOF), and multivariate Gaussian distribution (MGD) is proposed. To automatically cluster and annotate a specific user’s activity data, an improved K-Means algorithm with a novel initialization method is designed. By quantifying the samples’ informative degree in a labeled individual dataset, the most profitable samples can be selected for activity recognition model adaption. Through experiments, we conclude that our proposed models can adapt to new users with good recognition performance.

False Positive RFID Detection Using Classification Models

“…Integrating outlier detection (IQR or LOF) with the classification model improved classification accuracy in our dataset. Previous studies also showed that by integrating IQR or LOF with the prediction model, it improved the accuracy [29,33]. However, this presented model (RF with IQR or LOF) needs to be performed on different types of datasets, as it might not produce the highest accuracy compared to other models.…”

“…Experimental results showed that incorporating LOF reduced missed detections and the proposed method outperformed other considered methods. Zhao et al (2018) proposed human activity recognition by combining k-means, LOF, and multivariate gaussian distribution models [29]. LOF was used to remove outliers with low relative density from each cluster.…”

“…After cleaning the data (i.e., removing the identified outliers), several machine learning algorithms were applied and compared. gaussian distribution models [29]. LOF was used to remove outliers with low relative density from each cluster.…”

“…Collected data commonly includes outliers, due to environmental interference and imperfect sensing, and various outlier detection methods have been proposed to detect and remove them. The local outlier factor (LOF) has been applied for several application areas to detect outliers in data streams [25], fault detection in medical [26] and process monitoring [27,28], and activity recognition [29]. Interquartile range (IQR) has shown significant outlier detection capability for residential building energy usage [30], groundwater data [31], and health care insurance fraud [32].…”

A Multi-Mode PDR Perception and Positioning System Assisted by Map Matching and Particle Filtering

“…Unfortunately, there is a significant computational load associated with continuous sensor data processing, particularly in previous research [38][39][40][41] on pedestrian mode recognition.…”