Human activity recognition with AutoML using smartphone radio data

Abstract

Participants of the fourth edition of SHL recognition challenge 2021 aim to recognize eight locomotion and transportation activities in a user-independent manner based on radio data, including GPS reception, GPS location, WiFi reception, and GSM cell tower scans. Team "DD" proposes applying Google's AutoML Tables service to preprocess data, train, and evaluate the model. During this challenge, we showed the advantages and disadvantages of AutoML Tables. In addition, we have employed additional complimentary publically available datasets. AutoML Tables helped to train an artificial neural network model using the AdaNet algorithm. As a result, it has shown an ability to recognize classes with a precision of 81.2% and recall of 78.2%. Also, we opened the source code of the required feature engineering and published it on GitHub: https://github.com/dbalabka/shl-activity-recognition-2021. CCS CONCEPTS• Human-centered computing → Ubiquitous and mobile computing; • Computing methodologies → Machine learning; Machine learning approaches; Neural networks.

Accuracy Improvement of Vehicle Recognition by Using Smart Device Sensors

“…In Refs. [14][15][16], the best accuracy obtained was 93.4%, which was lower than our results, while also using more sensors than ours to detect mode of transportation.…”

“…Using Google's AutoML Tables service to preprocess and evaluate the Sussex-Huawei location transportation (SHL) dataset, Ref. [14] created 975 features by hand in order to achieve reasonable prediction quality. Smartphone GPS location data, GPS reception, Wi-Fi reception, and GSM cell tower scans were analyzed to determine whether the mode of transportation was standing still, walking, running, biking, driving a car, riding a bus, taking a train, or a subway.…”

“…Figures [14][15][16] show the preprocessing of the data, as it is initially input in Figure 14, and then passes through the following steps 1 and 2 respectively:…”

“…The raw dataset for each class has different lengths. As such, all of them are trimmed Figures [14][15][16] show the preprocessing of the data, as it is initially input in Figure 14, and then passes through the following steps 1 and 2 respectively:…”

Deep CNN-BiLSTM Model for Transportation Mode Detection Using Smartphone Accelerometer and Magnetometer

“…The sensors utilized in the acquisition of location-based data include GPS position, GPS reception, WiFi, and cellular technology. The studies conducted by [16], [17] have yielded favorable outcomes when employing the aforementioned four sensors. GNSS is another option to be used [18].…”