Keynote Speaker

Prof. Konstantinos Tserpes

"Maritime trajectory classification and clustering: problem definition and solutions"

Abstract:
The primary focus of this presentation is on leveraging Convolutional Neural Networks (CNNs) to extract meaningful features and patterns from vessel trajectories. A novel trajectory classification approach is proposed where trajectories are segmented and pertinent features such as acceleration, angular speed, and rate of change of course over ground are computed. These features are then quantified by their frequency of occurrence and normalized to construct a matrix representation. This matrix is subsequently fed into a CNN for classification, achieving high-accuracy results. Additionally, the proposed methodology is compared against a state-of-the-art technique that transforms trajectories into images, demonstrating its efficacy. Regarding trajectory clustering, an overview of trajectory clustering methodologies and evaluation metrics is given. Based on this overview, a CNN-based methodology is introduced. The matrix representation derived from the classification phase serves as input to a CNN. The loss function of the CNN is modified to optimize clustering metrics, including silhouette score, effectively exploiting the CNN's training process as a clustering approach.

Bio:
Konstantinos Tserpes is an Associate Professor in the Department of Informatics and Telematics at Harokopio University of Athens. He obtained his PhD in Distributed Systems from the School of Electrical and Computer Engineering at the National Technical University of Athens. His research interests focus on efficient computing and data analytics systems. In the field of spatiotemporal analysis he has worked extensively with his team in the development of applications using AIS vessel data and machine learning algorithms. He has actively participated in numerous EU and National funded projects and held roles as the scientific or general coordinator.