GRS-34806 Deep Learning in Data Science

Vak

Studiepunten 6.00

Language of instruction:

English

Assumed knowledge on:

FTE-35306 Machine learning and one Python programming course (e.g. INF-22306 programming in python or GRS-33806 Geoscripting or BIF-50806 Practical computing for biologists BIF-50806)

Contents:

Note: This course has a maximum number of participants.
The deadline for registration is one week earlier than usual. See Academic Year: https://www.wur.nl/en/Education-Programmes/Current-Students/Agenda-Calendar-Academic-Year.htm->Registration for Courses

This course offers an introduction to deep learning algorithms, with emphasis on their applications in life, earth and social science on image, sequence and numeric data. First, the mathematical basis of neural networks will be presented, and then a series of deep learning models will be discussed to solve problems of image segmentation, object detection, representation learning and sequence-based prediction. Every block of the course will include both a theoretical and an applied part, where students will implement their own solutions to solve a specific problem. Implementations will be done in Python. At the end of the course, an applied project in the fields of interest of life, earth or social sciences will be performed by the student as a final assignment. The course is a part of a sequence of courses offered by WUR for students who want to master their knowledge and skills on the relevance of data science for life sciences.  

Learning outcomes:

After successful completion of this course students are expected to be able to:
- explain basic concepts and mathematics underlying neural networks;
- understand and discuss scientific papers in the field of deep learning;
- describe characteristics and typical applications of different neural networks architectures and deep learning algorithms;
- propose deep learning approaches to solve problems in life, earth and social sciences;
- implement and evaluate effective deep learning solutions based on chosen algorithms to solve practical problems.

Activities:

- lectures;
- tutorials with hands-in coding exercises;
- practicals as computer labs;
- self study (including paper reading).

Examination:

- a group project during weeks 5 and 6 (50% of the mark);
- a written, closed book examination during the final week (50%).
Each mark will need to be higher than 5.5 to pass the course.

Literature:

Sandro Skansi. Introduction to Deep Learning: from logical calculus to artificial intelligence. Springer 2018.