Machine Learning for Earth and Environmental Sciences

In this 12-week practical course, we will introduce common machine learning algorithms in the context of their applications in Earth and environmental sciences. By the end of this course, students are able to identify common algorithms and summarize their advantages and limitations, especially in the context of environmental science, and implement them in Python (mostly using the Numpy, Scikit-Learn, Keras, and Tensorflow libraries in Google Collab notebooks). Meanwhile, aiming at the environmental application that they are passionate about (e.g., their thesis and further research), they will have the ability to choose appropriate algorithms from their existing experience.

These machine learning algorithms include but are not limited to:

• Linear/Logistic Regression for classification

• Decision Trees, Random Forests, Support vector machines

• Unsupervised Machine Learning for clustering

• Artificial Neural Networks, Convolutional Neural Networks, Recurrent Neural Networks, Graph Neural Networks

• Explainable Artificial Intelligence, such as Permutation tests, Partial-dependence plots, Saliency maps, Feature visualization, etc.

• Generative modeling, such as Auto-encoders, Generative adversarial networks, etc.