Learning Activity Predictors from Sensor Data: Algorithms, Evaluation, and Applications

Bryan David Minor; Janardhan Rao Doppa; Diane J Cook

doi:10.1109/TKDE.2017.2750669

Back

Learning Activity Predictors from Sensor Data: Algorithms, Evaluation, and Applications

Journal article

Open access

Learning Activity Predictors from Sensor Data: Algorithms, Evaluation, and Applications

Bryan David Minor, Janardhan Rao Doppa and Diane J Cook

IEEE transactions on knowledge and data engineering, Vol.29(12), pp.2744-2757

12/01/2017

DOI: https://doi.org/10.1109/TKDE.2017.2750669

Handle:

https://hdl.handle.net/2376/118318

PMCID: PMC5813841

PMID: 29456436

Files and links (1)

url

https://doi.org/10.1109/TKDE.2017.2750669View

Published (Version of record) Open

Abstract

Measurement

Training data

Smart homes

digital prompting

Activity recognition

Activity prediction

Prediction algorithms

Feature extraction

Environmental factors

smart environments

regression learning

Recent progress in Internet of Things (IoT) platforms has allowed us to collect large amounts of sensing data. However, there are significant challenges in converting this large-scale sensing data into decisions for real-world applications. Motivated by applications like health monitoring and intervention and home automation we consider a novel problem called Activity Prediction, where the goal is to predict future activity occurrence times from sensor data. In this paper, we make three main contributions. First, we formulate and solve the activity prediction problem in the framework of imitation learning and reduce it to a simple regression learning problem. This approach allows us to leverage powerful regression learners that can reason about the relational structure of the problem with negligible computational overhead. Second, we present several metrics to evaluate activity predictors in the context of real-world applications. Third, we evaluate our approach using real sensor data collected from 24 smart home testbeds. We also embed the learned predictor into a mobile-device-based activity prompter and evaluate the app for nine participants living in smart homes. Our results indicate that our activity predictor performs better than the baseline methods, and offers a simple approach for predicting activities from sensor data.

Metrics

9 Record Views

Details

Title: Learning Activity Predictors from Sensor Data: Algorithms, Evaluation, and Applications
Creators: Bryan David Minor - School of Electrical Engineering and Computer Science, Washington State University, Pullman, WA
Janardhan Rao Doppa - School of Electrical Engineering and Computer Science, Washington State University, Pullman, WA
Diane J Cook - School of Electrical Engineering and Computer Science, Washington State University, Pullman, WA
Publication Details: IEEE transactions on knowledge and data engineering, Vol.29(12), pp.2744-2757
Academic Unit: Electrical Engineering and Computer Science, School of
Publisher: IEEE
Grant note: R01EB015853 / National Institute of Biomedical Imaging and Bioengineering (10.13039/100000070) 0900781; 1262814 / US National Science Foundation (10.13039/100000001)
Identifiers: 99900547839801842
Language: English
Resource Type: Journal article