This article is part of a series of articles on different European master's programs related to artificial intelligence and machine learning.
The purpose of this post is to give an overview of Machine Learning related Master’s programs at ETH Zurich. Machine Learning is a topic that is researched and taught in different departments and institutes at ETH, which means that there exist many resources that can be hard to navigate. Our goal is to give you an inter-departmental overview of the Machine Learning landscape at ETH, and point you to the right resources. We also want to put an emphasis on opportunities that ETH provides for people interested in technical AI safety. This post is structured in two parts. The first part covers general information about ETH, its Master programs and its research possibilities in ML. The second part describes a few relevant Master’s programs in more detail.
ETH Zurich is consistently ranked as a top university: for example 14th overall and 3rd in Computer Science in the 2021 Times Higher Education Rankings, 6th in the QS World University Ranking, and currently 29th on csrankings.org when considering publications in the area of AI & ML.
Interestingly there are many different kinds of Machine Learning related Master programs that ETH offers. Most of them are 4 semester (120 ECTS) programs, with 30 ECTS assigned to the thesis. The Data Science Master program is most specifically tailored to an ML curriculum. However, admissions for this program are very competitive. Note that most courses of the Data Science Master can also be taken in other programs, such as the Computer Science Master, which allows students to adapt big parts of the curriculum to their interests.
There are numerous groups at ETH doing ML research. As ML is becoming more accessible for research in other fields, there also are many new groups using ML for their domain-specific applications. Consequently, there are countless courses on ML and related topics, offered by various groups. To give you a better idea of what kinds of subjects there exist, here are a few courses to give you an overall impression (note that this is by far an incomplete list):
- Introduction to ML
- Advanced ML
- Deep Learning
- Statistical Learning Theory
- Machine Perception
- Natural Language Processing
- Reliable and Interpretable Artificial Intelligence
- Probabilistic Artificial Intelligence
- Fairness, Explainability and Accountability for ML
- and many more...
As mentioned before there are also ML subjects that are more domain specific such as “Machine learning for Health Care” or “Deep Learning for Autonomous Driving”. Lastly there are countless mathematics and statistics subjects that are very related to ML, such as “Computational Statistics”, “Mathematics of Data Science” or “Optimization for Data Science”. You can find all the courses offered by ETH here. Many of the courses have mandatory projects that count toward the overall grade, which are great to get hands-on experience of the taught material.
Lastly we want to point out other opportunities that ETH offers. While this post mainly discusses ETH’s Master programs, doing a PhD at ETH can also be a great option. As mentioned previously there are numerous groups doing research on Machine Learning and its countless applications. Because these groups are growing rapidly, they often hire PhD students all year round. One perk of doing the doctoral studies at ETH, is that the university pays very competitive salaries (see here).
Furthermore, ETH also supports people who want to found a startup and apply their knowledge to create products on the market through the ETH Spin Off Program. This list of ETH Spin-offs shows that there are many well known companies that have come out of ETH.
ETH has many groups doing Machine Learning research, many of which are focused on a specific area of ML and others who use it for related fields. Below we give you a non-exclusive list of groups and research directions that exist at ETH, but note that there might be many more, as Machine Learning is becoming increasingly used in other fields as well.
- Institute for Machine Learning
- Computational genetics and epigenetics of cancer (Prof. Valentina Boeva)
- Information Science and Engineering (Prof. Joachim Buhmann)
- Data Analytics (Prof. Thomas Hofmann)
- Learning and Adaptive Systems (Prof. Andreas Krause)
- Biomedical Informatics (Prof. Gunnar Rätsch)
- Medical Data Science (Prof. Julia Vogt)
- Statistical Machine Learning (Prof. Fanny Yang)
- Institute for Visual Computing & Institute for Pervasive Computing
- Other groups
- Institute of Neuroinformatics
There is no “AI Safety” group at ETH, but many individuals in different groups do things related to it, e.g., in the Learning and Adaptive Systems group and in the Secure, Reliable, and Intelligent Systems Lab.
Life outside of University
Zurich is consistently ranked as one of the top ten cities with the highest living quality (e.g. some ranking in 2019). The city is located at lake Zurich, it is surrounded by many hills, lots of nature and is close to the Swiss Alps. Everyone who loves the outdoors will definitely not be disappointed as there are so many things one can do. The city is vibrant and there are many cultural activities and a great nightlife one can pursue. The old town has a lot of charm and many restaurants and bars where one can pass a nice evening. Unfortunately, Zurich also is consistently ranked as one of the most expensive cities in the world (see for example this article). If you work in Zurich, this is usually not a problem, because salaries are also high. But if you come from a foreign country that has lower living costs, this might be a problem. ETH Zurich itself does offer different kinds of grants for people who can’t afford to study on their own (see here), but they might be quite competitive. Another important thing to note is that in Switzerland, universities usually do not have a campus where students can live (such as for example in the US). This means that students usually have to look for “normal” apartments that they often share with flat mates. There are however organisations that try to provide affordable renting for students (see Juwo, note that according to their website they should be open for foreigners. But as we have no experience with them, we can’t guarantee anything). ETH also provides a few housing opportunities for students (see this link for more information on searching for housing and opportunities of ETH).
If you are a student at ETH Zurich you will have access to the ASVZ. The ASVZ is the association that incorporates all sport activities for students. That means you will have access to several gyms, group fitness and any kind of sport you can imagine such as Soccer, Volleyball, Badminton etc. This is a great opportunity to stay fit and meet people. There are also other student associations such as bands, model united nations and many more. Often there is also an association for the department that you belong to that organizes fun events, talks by companies etc. We want to note however, that there are probably much fewer associations by the University than in other countries. In Switzerland it is common for people to also join external associations such as soccer clubs, choirs etc. For people interested in AI Safety we also recommend Effective Altruism Zurich. It is a group of people with various backgrounds that organize events, discussions and projects. It also has a recurring AI Alignment discussion group in which recent research on AI Safety is discussed.
Finally, we want to point out that there are quite a few opportunities for people that have a degree in Machine Learning. Due to the economic situation of Zurich and its high living standard, many tech companies have settled here. Google for example has its Europe headquarters in Zurich. Also IBM, Microsoft, Apple, Disney and many more do ML research in Zurich. However, other places such as the Silicon Valley will have a more vibrant tech community (note that we are saying this by pure intuition, we don’t base this on any actual facts).
Specific Master’s Programs
Master in Computer Science
The masters program in Computer Science (CS) at ETH (see also Study Guide), has been completely reorganized in 2020. What used to be a 90 ECTS Master, has now been extended to 120 ECTS (similarly to the Master In Data Science) and has gotten a new structure of major and minor tracks (We want to point out we only have personal experience with the “older” CS Master’s program that has a different structure and only 90 ECTS. Thus we can’t really give any insights about the whole experience of the new Master’s structure etc. Also even though we try to be diligent about the information, take all specific advice with respect to the new Master with a grain of salt.). The CS Master track provides 5 different Major directions, such as Data Management Systems, Machine Intelligence, Secure and Reliable Systems, Visual and Interactive Computing and Theoretical Computer Science. Additionally there are minor directions that should give the student a broader curriculum, including other aspects of Computer Science. For a given Major, one can only choose Minor tracks that cover a different area of Computer science (see image below). You will also have to choose a tutor i.e. a professor, who will help you with the selection of your courses and help you plan your whole Master. The program will be completed with your masters thesis that will count towards 30 credit points.
(image taken from https://inf.ethz.ch/studies/master/master-cs-2020.html)
Obviously for someone interested in Machine Learning we would recommend the Machine Intelligence Major track. After choosing a Major and Minor track, one has the possibility to fill up all the required credits for each category with courses that fall within these categories (see image below). That means one still has a lot of possibilities in the choice of concrete courses. Additionally one can also take Free Elective Courses which are all master’s level courses offered by ETH Zurich, EPF Lausanne and the University of Zurich. Even though this is only a small part of your course portfolio, it is the perfect opportunity to take some subjects that really interest you but might have nothing to do with Computer Science. One can also see in the image below that it is required to take courses within the categories of Practical Work and Inter Focus Courses. Practical Work is a really cool category that allows you to get hands-on research experience. To complete this category, you need to either do a semester project in a research group or take a lab course. The Inter Focus Courses are (at least in the old Master structure) lab courses where a big part of the semester is dedicated to practical projects. Overall I believe that this Master’s structure is highly beneficial to a person interested in ML/AI, as it can provide you with the necessary theoretical and practical knowledge required in this field and additionally give you the background of other highly related fields in Computer Science.
(image taken from https://inf.ethz.ch/studies/master/master-cs-2020.html)
Master in Data Science
The masters program in Data Science (DS) at ETH is a joint offering from the computer science, mathematics and electrical engineering departments. The core topics of the program are statistics, machine learning, optimization and data management. While there are some core courses in these topics, you are generally relatively flexible in selecting courses and a direction to specialize in. In addition to the core courses and electives, you have to select an interdisciplinary elective, which is an area of research in which machine learning can be applied. There are a lot of options such as Computational Biology, Social Networks, Transportation or Weather and Climate. Similar to the computer science program, every student can choose a tutor (a professor affiliated with the program), who can help with selecting courses and planning your studies.
The Data Science masters provides a unique opportunity to specialize in modern machine learning techniques during the masters, because there are no other required courses that are not directly relevant to this topic. Also, it can be a great opportunity for strong students with non-CS backgrounds, because the entry requirements are quite flexible (a lot of students have bachelor degrees in subjects such as Physics, electrical or mechanical engineering).
Similar to the CS masters, completing the DS masters also requires 120 credit points, 30 of which are a masters thesis. Additionally there is the 14 credit Data Science lab, which is unique to this program. It is essentially a one-semester research project with researchers at ETH or external companies. The projects are usually real research questions or applications of machine learning in the real world. So this is a great opportunity to gain experience in practical projects. With the data science lab, the master thesis and optional semester projects the DS masters offers a lot of opportunities to get research experience to prepare for a PhD or industry positions.
Master in Neural Systems and Computation (NSC)
INI was founded in 1995 as a joint institute of University of Zurich and ETH Zurich. Many researchers at INI come from Caltech, and the NSC programme mirrors its sister master’s programme Computation and Neural Systems master at Caltech (started by Carver Mead, John Hopfield and Richard Feynman).
The research at INI can be broadly separated into systems neuroscience, theoretical neuroscience and neuromorphic engineering. While the institute was founded by pioneers of systems neuroscience, its greatest strength today lies in neuromorphic engineering. Due to the recent global resurgence of interest in machine learning and deep learning, there is increased research in these areas (see the Research Directions section above).
However, the general attitude at INI is that machine learning is useful but hyped, and the really interesting questions are how the brain does computation and how we can build fundamentally different brain-inspired technology. If you are only interested in machine learning and not neuroscience, the NSC is probably not for you.
Undertaking the NSC means becoming part of INI. As INI is a proud institute with historical ties to scientific pioneers, you are in some sense joining a family / community with a legacy. INI has its own unique culture that values creative spaces, and you are expected to attend several weekly events (lab meeting, journal club, colloquium / happy hour). As a member of INI, you will be responsible for co-organizing the journal club and happy hour. NSC students additionally organize monthly aperos where we invite academic speakers.
You have a lot of freedom in the NSC programme. There are some mandatory courses and some semi-mandatory courses (you must obtain a certain amount of course credit points from 2 out the 3 course groups which are systems neuroscience, theoretical neuroscience and neurotechnology). However, beyond that you are essentially free to take any courses offered by University of Zurich and ETH Zurich. You are allowed 3 years to finish the degree, and it is entirely your responsibility to graduate within time. For most people, the degree should be possible to complete in around 1.5-2 years, but some students have managed to do it in less. Details about the program structure can be found here.
The NSC is very research focused, and you can do either 3 short projects (2x1.5 months mini-projects and 1x6 months master’s project) or 1 long project (1x9 months master’s project). Your master’s project is expected to be of publishable quality and you are encouraged to publish, but it is not a requirement. Permissions can be granted to do projects externally, but generally you are encouraged to do projects at INI.
- Admission is competitive and the admission rate is rumored to be around 10-20%.
- You select 2 potential mentors from the institute as part of the application process, and you will most likely be interviewed by one of these if you pass the initial round.
- Personality matters. You will not be admitted without strong passion for neuroscience or neuroscience-inspired technology.