GEOG788T

Introduction to deep learning and its uses in spatio-temporal problems and applications. A variety of network types will be discussed, including densely-connected networks, convolutional neural networks, graph convolutional networks, recurrent and gated networks, generative adversarial networks, Siamese networks, etc. The discussion will also include related topics on meta-learning, reinforcement learning, spatial-explicit frameworks, knowledge-guided formulations, etc. The techniques will be discussed in the context of spatial and spatio-temporal data for applications including Earth observation (e.g.,change detection, land-cover classification, geospatial object detection), COVID-19 mobility estimation, traffic accident prediction, and more. The course will also cover implementation using Python. Students will carry out projects based on their own research topics or interests, and the projects can be either applied (existing methods on new data) or technically innovative (improving existing methods).