Using deep learning for trajectory classification in imbalanced dataset

3Citations
Citations of this article
5Readers
Mendeley users who have this article in their library.

Abstract

Deep learning has gained much popularity in the past years due to GPU advancements, cloud computing improvements, and its supremacy, considering the accuracy results when trained on massive datasets. As with machine learning, deep learning models may experience low performance when handled with imbalanced datasets. In this paper, we focus on the trajectory classification problem, and we examine deep learning techniques for coping with imbalanced class data. We extend a deep learning model, called DeepeST (Deep Learning for Sub-Trajectory classification), to predict the class or label for sub-trajectories from imbalanced datasets. DeepeST is the first deep learning model for trajectory classification that provides approaches for coping with imbalanced dataset problems from the authors’ knowledge. In this paper, we perform the experiments with three real datasets from LBSN (Location-Based Social Network) trajectories to identify who is the user of a sub-trajectory (similar to the Trajectory-User Linking problem). We show that DeepeST outperforms other deep learning approaches from state-of-the-art concerning the accuracy, precision, recall, and F1-score.

Cite

CITATION STYLE

APA

Freitas, N., Silva, T., Macêdo, J., & Junior, L. (2021). Using deep learning for trajectory classification in imbalanced dataset. In Proceedings of the International Florida Artificial Intelligence Research Society Conference, FLAIRS (Vol. 34). Florida Online Journals, University of Florida. https://doi.org/10.32473/flairs.v34i1.128368

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free