Enhancing user identification through batch averaging of independent window subsequences using smartphone and wearable data
Enhancing user identification through batch averaging of independent window subsequences using smartphone and wearable data
By
Rouhollah Ahmadian, Mehdi Ghatee, Johan Wahlström
has been accepted in “Computers & Security“.
Highlights:
- •Introduced primary and mixed patterns for spatio-temporal data classification.
- •Addressed uncertainty due to pattern similarity and random primary pattern mixing.
- •Proposed the Batch Averaging Probabilities (BAP) model for robust classification.
- •BAP segments and classifies subsequences, averaging predictions to improve accuracy.
- •Theoretical proofs show BAP reduces error variance and improves performance.
Abstract
Throughout daily life, individuals partake in various activities such as walking, sitting, and drinking, often in a random manner. These physical activities generally exhibit similar patterns across different people, posing a challenge for identifying users using smartphone and wearable data. To tackle this issue, we have developed a new model called Batch Averaging Probabilities (BAP). Our approach involves segmenting input sequences into separate windows, independently classifying each segment, and then averaging the probabilistic predictions to make the final decision. The BAP method introduces the concept of primary patterns, which are the smallest meaningful sequences. It effectively deals with the random order of primary patterns within mixed patterns. Our work includes theoretical evidence supporting the BAP method, showcasing its ability to minimize prediction variance and enhance model accuracy. Additionally, the model’s training algorithm employs a unique approach. Model selection and regularization are based on the averaged loss of segments, reducing overfitting and improving performance without the complexity associated with using an ensemble of neural network models. We evaluated the effectiveness of our proposed method using accelerometer and gyroscope data from diverse user activity datasets including UIFW, WISM, HOP, CLD, RSSI, DI, DB2 and HAR, demonstrating significant performance improvements over state-of-the-art models. Specifically, our approach outperforms DB2 by 1.08%, HAR by 7.67%, and DI by 14.76% in terms of accuracy.
Ref: https://doi.org/10.1016/j.cose.2024.104265
The full version of the paper “Enhancing user identification through batch averaging of independent window subsequences using smartphone and wearable data” by Rouhollah Ahmadian, Mehdi Ghatee, & Johan Wahlström is now available online in “Computers & Security”. It can be downloaded by clicking the following Share Link before February 04, 2025. No sign up, registration, or fees are required.