Activity Recognition with Evolving Data Streams

Abstract

Activity recognition aims to provide accurate and opportune information on people's activities by leveraging sensory data available in today's sensory rich environments. Nowadays, activity recognition has become an emerging field in the areas of pervasive and ubiquitous computing. A typical activity recognition technique processes data streams that evolve from sensing platforms such as mobile sensors, on body sensors, and/or ambient sensors. This paper surveys the two overlapped areas of research of activity recognition and data stream mining. The perspective of this paper is to review the adaptation capabilities of activity recognition techniques in streaming environment. Categories of techniques are identified based on different features in both data streams and activity recognition. The pros and cons of the algorithms in each category are analysed and the possible directions of future research are indicated. CCS Concepts: r General and reference → Surveys and overviews; r Theory of computation → Streaming models; Machine learning theory; r Computing methodologies → Transfer learning; r Hardware → Sensor applications and deployments; Sensor devices and platforms;

C2FHAR: Coarse-to-Fine Human Activity Recognition With Behavioral Context Modeling Using Smart Inertial Sensors

“…These steps include: 1) data acquisition, 2) data preprocessing, Proposed methodology for C2FHAR based on multi-label classification model. Here, c (1) and c (2) represent the predicted class labels for coarse-level and fine-level HAR, respectively, whereas p (1) and p (2) represent the predicted probabilities of c (1) and c (2) respectively.…”

“…The associate editor coordinating the review of this manuscript and approving it for publication was Benyun Shi . information as regards to the people's activities by processing data streams originating from different sensing modalities (such as video cameras, wearable/on-body inertial sensors, mobile sensors, and/or ambient sensors) [2]. In recent years, sensor-based HAR has become increasingly significant in wide-ranging disciplines, including human-computer interface (HCI), ambient assistive living in smart homes [3], [4], driving behavior analysis [5], [6], robotics [7], [8], and telecare for personal health monitoring [9].…”

ProCAVIAR: Hybrid Data-Driven and Probabilistic Knowledge-Based Activity Recognition

“…The major drawback of those approaches is that they require the acquisition of a large amount of labeled data to initialize the recognition model. For this reason, semi-supervised learning is emerging as a powerful tool to initialize the recognition model with few labeled data points and to continuously improve it over time [3], [11]- [13]. Among the many semi-supervised techniques (e.g., self-learning, co-training), active learning is one of the most effective [14]- [17].…”

A Taxonomy for Combining Activity Recognition and Process Discovery in Industrial Environments

“…Activity recognition (ARC) seeks to accurately identify human activities on various levels of granularity by using sensor readings. In recent years ARC has become an emerging field due to the availability of large amount of data generated by pervasive and ubiquitous computing [2,4,18]. Methods have demonstrated an increased efficiency in extracting and learning to recognise activities in the supervised learning setting using a range of machine learning techniques.…”

“…Assigning the correct ground truth label is a very timeconsuming task. There has been less work on unsupervised [21] or semi-supervised techniques [19] which require fewer annotations [2,39]. Another challenge lies in the emergent topic of transfer learning [29], which helps with the redeployment of an ARC model from one factory floor to another with a different layout, environmental factors, population and activities.…”