A Multi-task Deep Learning Architecture for Maritime Surveillance using AIS Data Streams
Texte intégral
Documents relatifs
Considering the effect of supervised scenario, we show that from 50% of the original test set size, we surpass state-of-the-art performance on Avenue, Ped1, and ShanghaiTech..
We use the statistical features of multi-flows rather than a single flow or the features extracted from log as the input to obtain temporal cor- relation between flows, and add
In this work, we propose a multi-task model which explores deep learning, and more specifically recurrent neural networks to process AIS data stream for multiple purposes:
Our method provides a supervised learning solution to classify an unknown column in such tables into predefined column classes which can also come from a knowledge graph..
While in some works a secondary model was used as a source of contextual information flowing into the object detector, our design will attempt to reverse this flow of
Unsuper- vised reconstruction of sea surface currents from AIS maritime traffic data using learnable variational models.. ICASSP 2021: IEEE International Conference on Acoustics,
Generally worse than unsupervised pre-training but better than ordinary training of a deep neural network (Bengio et al... Supervised Fine-Tuning
The results obtained by γ-PhysNet on Monte Carlo simulations show that full-event reconstruction form a single IACT data is possible with a deep multi-task architecture.. In a