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

Abstract

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 9 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.

Smart home-driven digital memory notebook support of activity self-management for older adults

“…The activity forecasting algorithm was validated in the context of activity prompting for nine homes with a number of activities ranging from 2 to 9 [29]. In this evaluation we observed a range-normalized mean absolute error of 0.02 seconds.…”

“…The EMA app occasionally queries the resident about their current activity and we use the response as ground truth to measure performance of AR. Based on this validation mechanism for 8 users for 11 activities, we observed a 86% accuracy [29]. TM CO EA WD RE LH EH WO BT 3224 1 4 0 0 0 0 0 0 0 0 0 SL 1 Activity Prediction The DMN app also partners with the smart home to predict when activities or tasks should be completed.…”

Easing Power Consumption of Wearable Activity Monitoring with Change Point Detection

“…While CPAM can provide activity-aware energy reduction for many mobile applications, here, we focus on an activity recognition application. Human activity recognition is a popular research topic [32][33][34][35] and forms a critical component of technologies for health monitoring, intervention, and activity-aware service provisioning [36][37][38]. Additionally, activity recognition provides a vehicle for us to validate our change point detection methods by comparing detected activities with known activity transitions.…”

A Multi-task Learning Model for Daily Activity Forecast in Smart Home

“…Mahmud et al forecasted the next daily activity occurrence time based on the Poisson process [29]. Similarly, Minor et al independently trained a predictive regression model for occurrence time of specified daily activities based on additional feature sets [14,30]. Due to their accessibility limitations, it was not always feasible to add additional feature sets to the model.…”