Organizers
- Esther Rodrigo Bonet, University of Valencia and Vrije Universiteit Brussel,Belgium
- Paolo Frazzetto, University of Valencia, Spain
Abstract
Sensing devices and processing methods, recently including machine-learning techniques, have played a critical role in the modelling of the Earth system. Yet, the Earth encompasses complex processes which are challenging to capture and address. While the domains of Earth Science, Machine learning and Signal Processing have tackled them from different perspectives, the intersection of these domains has been achieving state of the art results while posing new interesting challenges. The session welcomes advancements in artificial intelligence, remote sensing technologies, signal processing techniques, and machine learning methodologies for addressing such critical Earth-related challenges. Topics include innovative approaches for (distributed) remote and on-site sensing and monitoring, environmental and climate data analysis (anomaly and extreme event detection, disaster, urban and natural resources assessment, causal discovery…) and societal and urban modelling (including stressors such as electromagnetic radiation, weather, traffic, air pollution or noise).
List of Topics Covered in the Special Session
The topic lays at the intersection between machine learning and signal processing. The session aims to bring together researchers to showcase novel solutions and its integration to enhance our understanding of Earth’s systems. The topic poses a direct impact and further pushes the application domain on Earth science, including:
- New domain-specific (distributed) remote sensing settings and techniques
- Innovative design of domain-specific signal processing and deep learning models
- New deep learning architectures, including graph deep learning models, transformers, generative deep learning, and their applications to Earth systems
- Discovery of causal relations and governing equations of Earth systems
- Out-of-distribution/Extreme event detection with deep learning
- Unsupervised, self-supervised and semi-supervised deep learning for the Earth systems
- Explainability and interpretability of deep learning methods
- Distributed and federated model-aware deep learning
- Applications in image/video sensing and processing for Earth systems, including machine and deep learning, signal processing, computer vision, big data, and natural language processing for Earth processes (environmental, climate and urban settings).