Deep reinforcement learning is cutting-edge method in AI. It is a combination of deep learning and reinforcement learning, which has led to AlphaGo beating a world champion, which can play Atari games at a superhuman level, and which has been applied in self-driving cars.
In this course, the students will learn the basic theory in reinforcement learning and implement algorithms of many deep reinforcement learning models in Tensorflow and OpenAI gym environment.
We use the Python programming language for the entire course. While using various open source libraries for computing and data visualization, we will also introduce the students the best practices in open science and how to contribute to open source projects. Lectures are devised to be interactive and to give the students enough time to acquire direct hands-on experience with the materials. Students will work in pairs throughout the tutorials and will team up to practice the newly learned skills in a real programming project — an entertaining computer game.
Learning Goals and Syllabus
In this course, the students will learn how to
● Apply a variety of advanced reinforcement learning algorithms to any problem
● Writing, Organizing, documenting, and distributing scientific code in Python
Please click here  for the course syllabus.
You may find the syllabus useful when discussing with your home University whether the ECTS credits attainable for this course are accepted by them.
Participants of the TU Berlin Summer University must meet the following requirements: (i) B2 level English, or equivalent and (ii) at least one year of university experience.
- Basic programming skills (students should be able to write and run small programs in the language of their choice and know what software library and interface means)
- Basic knowledge in linear algebra and statistics/probability theory, e.g., gradient, probability distributions
- Students should bring their own laptops to class
Dr. Vaios Laschos is a research associate in the Neural Information Processing Group of Technische Universität Berlin. He received his PhD in applied mathematics from Bath university in 2013, and since then he completed postdoctoral programs in various topics (PDEs, Control theory, Stochastic control theory, MDPs,). His main focus at the moment is MDPs and their application in RL.
Dr. Rong Guo is a research and teaching associate in the Neural Information Processing Group of Technische Universität Berlin. She received her PhD from the Faculty Electrical Engineering and Computer Science of TU Berlin in 2015. She works on both machine learning and neuroscience problems. Her focus is on reinforcement learning (deep or meta). She regularly teaches machine learning and programming to degree program students at the TU Berlin.
Msc. Youssef Kashef received the B.Sc. degree in electronics engineering from the American University in Cairo, New Cairo, Egypt, in 2010 and the M.Sc. degree in neural systems and computation from ETH Zürich, Zürich, Switzerland, in 2013. From 2012 to 2014, he worked as a Computer Vision Scientist at Affectiva, Inc. He is currently a graduate Research Assistant in the Neural Information Processing Group at Technische Universität Berlin, Berlin, Germany. His research interests include deep learning, machine learning with applications in machine vision and audition.
Please direct questions about the course to the TU Berlin Summer University Team at: summeruniversity(at)tubs.de . We will answer your questions and direct specific queries regarding course content to the course lecturers where necessary.