What is computer science (datalogy) in your opinion? We have asked this question to many current and former DIKU researchers and have quite a few different answers.

Computer science, Datalogi, and Informatik are for me the science of (systematic) information processing. I think this is a rather common definition.

What is the best introduction to your field of research?

The DIKU course "Statistical Methods for Machine Learning" provides a good introduction to my main field of research, machine learning. There are a couple of good textbooks on machine learning, for example:

• Bishop, C. M. (2008). Pattern Recognition & Machine Learning. Springer.
• Hastie, T., R Tibshirani, and J. Friedman (2001). The Elements of Statistical Learning: Data Mining, Inference & Prediction. Springer.
• Mohri, M., A. Rostamizadeh, and A. Talwalkar (2012). Foundations of Machine Learning. MIT Press.

For undergraduates, the free Stanford online course on machine learning is often recommended. I do not know the course in detail, but the instructor Andrew Ng is an excellent researcher and I heard a lot of positive comments on the course.

When did you write your first “Hello world!”?

I was very young, my older brother inspired me. I learned programming on a Commodore VC 20, so I am really into zero page addressing. My first program was in BASIC, which I learned from a book that introduced programming the Commodore PET.

What is the best program you have ever written? Or the one you are most proud of?

I am most proud on my contributions to the Shark machine learning library.

How did you end up at DIKU?

I was excited about the job offer I saw on a machine learning mailing list, applied - and here I am.

How do look at the future of DIKU?

I am very positive. However, I think that the mathematics education in our BSc programme must be improved, especially significantly extended. A solid mathematical foundation is important for almost all areas of computer science, for instance for the design and analysis of algorithms, the design and evaluation of experiments in human-centered computing, computer graphics, robotics, image analysis, artificial intelligence, etc. This must be reflected in the BSc education, as, for example, in the popular computer science BSc programme at Stanford.

Who is the best teacher you have had in your own CS education?

As a student, I had the pleasure of attending lectures by
Ingo Wegener and Helmut Thiele.

Helmut Thiele suddenly drawing a huge sword, which was hidden behind his desk, to explain Occam's razor in a lecture is something you do not forget quickly. Ingo Wegener impressed me a lot. I think that Ingo Wegener has been a role model for excellent teaching and research. Furthermore, I had the luck of doing my doctoral studies under the supervision of Werner von Seelen, whom I deeply admire. But he would not call himself a computer scientist.

If you weren't a computer scientist, what would you then be?

Well, I always wanted to be a guitarist, but unfortunately I am not good enough. If I had not studied computer science, I would most likely have studied biology or mathematics and would presumably be a researcher in one of these areas.

Which book or paper (not necessarily in computer science) do you wish everyone had read?

Herbert Marcuse's "Eros and civilization. A philosophical inquiry into Freud" - just kidding. A lot of great scientific papers and nice novels come to my mind, but I find it impossible to name a single paper or book. Brecht, Dürrematt, Hesse are never outdated, make you think and are fun to read - they are highly recommended. Well, I guess here my German mother tongue shines through.

What did you do before you came to DIKU?

I was Juniorprofessor at the Institute for Neural Computation at the Ruhr-University Bochum, Germany.

You teach the second part of Databases and Datamining. Is there any difference between the Computer Science students who take your course and the Nat-IT students?

The course Databases and Datamining has been designed for Nat-IT students. Due to several requests, we opened it for computer science BSc students, and I think that this was a good idea. I currently have no reliable statistics about the differences in performance between computer science and Nat-IT students. However, from the course evaluation questionnaire it seems that some of the computer science students perceived the mathematics to be difficult. This is a problem that must be addressed in different ways, starting by making the mathematical content more explicit in the course description.

Is there an area where you believe that machine learning is sorely under-used?

There are many such areas, for instance greentech.

Dikutal would like to thank Christian Igel for participating in this interview. We hope you found it interesting.