Day 2 Session 3: Deep Learning for Remote Sensing Data Analysis
Session 3:
Inverse problems (denoising, super-resolution)
Generative models (autoencoders and GANs)
Self-supervised learning
Overview:
This course will explore the application of deep learning techniques for remote sensing data analysis. The curriculum will include an introduction to machine learning, focusing on different learning architectures, problem types, data types, and challenges, as well as mathematical aspects such as optimization, regression vs. classification, objectives, and losses functions. Participants will gain an understanding of various deep learning architectures, including CNNs, RNNs, LSTMs, ConvLSTMs, and state-of-the-art transformers, which are becoming essential for the effective analysis of spatio-temporal data.
The course will cover both discriminative models for tasks like classification, detection, and regression in remote sensing, as well as generative models for image enhancement and forecasting. Furthermore, the application in decision-making will also be presented through topics such as Markov decision processes and deep reinforcement learning. Last, an overview of additional subjects will be presented including topics like uncertainty, expandability, physics-informed DNNs, and foundational models.
Instructor:
Grigorios Tsagkatakis, Associate Professor, University of Crete and Institute of Computer Science, and affiliated researcher, Foundation for Research and Technology - Hellas (FORTH)
Grigorios Tsagkatakis is an associate professor at the Computer Science Department of the University of Crete and an affiliated researcher at the Institute of Computer Science of the Foundation for Research and Technology – Hellas (FORTH) in Greece. He received his BE and MS degrees in Electronics and Computer Engineering from the Technical University of Crete, Greece in 2005 and 2007 respectively, and his Ph.D. in Imaging Science from the Rochester Institute of Technology, New York, in 2011. Between 2019 and 2021, he was a Marie Skłodowska–Curie fellow at the Department of Electrical and Computer Engineering of the University of Southern California with Prof. M. Moghaddam. His research focuses on topics related to signal/image processing and machine learning with applications in remote sensing and astrophysics.
Slides and codes can be found here:
https://github.com/gtsagkatakis/GRSS2...