Instructors: Dr. David Greenberg
Event type:
Seminar
Displayed in timetable as:
OZ-M-DL
Hours per week:
4
Credits:
6,0
Language of instruction:
English
Min. | Max. participants:
5 | 24
Comments/contents:
This course will prepare students to effectively use deep learning to solve classification and regression problems. Lectures will cover the conceptual and mathematical aspects of deep learning, while weekly programming exercises will provide hands-on experience in applying these ideas to data from the geosciences, using
Python, NumPy and Pytorch. This course aims to give students a broad hands-on competence in deep learning fundamentals, including optimization, autodiff, convolution, recurrence, self-attention, autoencoders and deep generative models. The course will conclude with individual projects to be presented in class.
Learning objectives:
Students will have understood fundamental neural network approaches to classification and regression problems. They will have written programs implementing multiple neural network architectures and trained them on simulations and observations of the atmosphere and ocean. They will have hands-on experience in designing and executing a deep learning-based research project.
|