How's it going with the new Master of Data Science and Artificial Intelligence?

Growing up with a new master's

January 13, 2022

When the new master's program in Data Science & Artificial Intelligence was just a twinkle in someone's eye, Dirk Fahland was there, keen to help nurture it to fruition. That was two and a half years ago. “I feels a little like it's my baby.”

Dirk Fahland. Photo: Loraine Bodewes

Pregnancy, birth and child raising - an accelerated process of creation and nurturing is how creating the new Master of Data Science and Artificial Intelligence felt for Associate Professor Dirk Fahland. With the inevitable teething troubles and growing pains. In September 2021, the master's program was launched, for a cohort of no fewer than 155 students. Fahland: “That number exceeded all our expectations and brings it own challenges. But equally, it is evidence of the huge demand there was for this program.”

It has, of course, given Dirk Fahland a few sleepless nights. The immense amount of work involved in creating and structuring the Master of Data Science & Artificial Intelligence (DS&AI) was one thing. The lengthy accreditation process, stretching to eighteen months, required for the program was another. The huge numbers of students who applied before the new master's was up and running; there was plenty to tax the mind.

And what about the jigsaw puzzle of staffing issues: how do you find lecturers to teach all the courses in the new master's and to supervise the master's students? The teaching load of lecturers at the Department of Mathematics & Computer Science was already sky high due to the teaching services they deliver to every other department. And in the relatively new, but booming field of data science and artificial engineering there's no surplus of qualified staff, eager to be snapped up. On the contrary, they haven't yet been trained.

Spiritual parent

In time, Fahland became the driving force behind the project group taking on the new master's, and was among the ‘spiritual parents’ on hand to help with ailments arising during the pregnancy, the delivery itself and the first childhood illnesses. “I must admit that I've come to regard the master's as a child of sorts. I want this to work for our students and the department. In time, the students we are educating must help society. And so I feel a huge responsibility for the program.”

I must admit that I've come to regard the master's as a child of sorts

Dirk Fahland, driving force Master of DS&AI

Photo: Loraine Bodewes

How it started: the missing master's

It was in February 2019 that the realization struck within the Department of Mathematics and Computer Science: something was missing. Students looking for a suitable follow-up program having completed the Bachelor's of Data Science found our university had nothing to offer them. Even though for the past ten years or more, research on data science had been conducted within the department, within various groups.

There was, however, a smoothly running track - Data Science in Engineering - within the Master of Computer Science and Engineering. Whose popularity and student cohort was starting to outstrip those of the master's itself. A significant signal that a follow-up step in this field was very much needed. “We were unable to offer our own Data Science bachelor's the option of completing their studies with a directly matching master's, or of going on to do doctoral research in the field,” says Fahland.

A project group was set up and Fahland lost no time in joining. “They were looking for someone keen to think about the content of the new master's. What should it include, what do we want our students to learn? Being the program manager for the Data Science in Engineering track, I knew at once that I wanted to be involved.”

How AI came on board

The remit was initially to design a Master of Data Science. “There was certainly a need for it, but during our early discussions it was clear that the field of artificial intelligence was becoming increasingly important both within our department and across the university,” says Fahland. In 2019 TU/e founded the Eindhoven Artificial Intelligence Systems Institute (EAISI), which brings all AI activities at TU/e together under one roof. Research, education and student teams. “The question soon being asked was ‘and what are we doing with AI?’ The board of our department seized the opportunity and decided we should add AI to the master's.” The plan for DS&AI was conceived.

The question soon being asked was ‘and what are we doing with artificial intelligence?’

Dirk Fahland, driving force Master of DS&AI

 

The new master was kicked off in September 2021. Photo: Bart van Overbeeke

He was not immediately enthused by the addition of AI, but since then Fahland has embraced it wholeheartedly. “We were creating a master's program in which we teach students to derive predictive models from data, a process in which they use techniques widely used in artificial intelligence. So actually it was entirely logically that we include AI in the new master's.”

“We are continuing to add new courses that are fundamental to a modern AI degree program. And in so doing we've given AI a proper place within our department,” believes Fahland. “While the combination of data science and artificial intelligence isn't common - in fact, we are the first to offer this in the Netherlands, and there are only a handful of programs worldwide - personally, I think it is a very good combination. The two fields are more intertwined than people think,” says Fahland.

He expounds: “Data science asks the question - and helps provide the answers - how do people use data when they are applying artificial intelligence. And when should you not be using your data because it isn't good enough. In turn, artificial intelligence can automate boring and repetitive tasks involved in data processing. This opens the way to retrieving so much more from data than a person could retrieve; we can get overwhelmed by a huge quantity of data.”

A few weeks later I found myself to be the driving force and in the thick of it

Dirk Fahland, driving force Master of DS&AI

Taking up the slack

Fall 2019, at the end of the first quartile. Within the project group the tasks were equally distributed, “After the initial sounding board sessions with students I had a good idea of how the master's should take shape. Then the second quartile started and the other lecturers in the project group had large courses to teach. Work pressure diminished their mental capacity to engage fully in our process. I took up the slack. It wasn't a case of me unilaterally deciding ‘we'll do it like this’; everyone embraced it. A few weeks later I found myself to be the driving force and in the thick of it.”

Fahland is grateful for the opportunity to work so closely on the new master's program. “When it comes to solving problems at our department, Mathematics and Computer Science, no one looks at rank or how long you have been working here. If you have something to contribute, you are on board. I think the way I went about contributing during the development process gave others the confidence that I could do it.”

Challenge-based learning

An ambitious timeline was put in place: writing a 'macro-efficiency test' as it is known, benchmarking against other programs, submitting the accreditation application and designing the curriculum of the new master's.

While the field continued to develop apace, the project team gave themselves two months to consider which courses and in what form would be essential to the program. “From the outset, I knew I wanted the master's to be challenge-based,” says Fahland, “just like the bachelor's. It is fantastic to see what our students manage to achieve with data when they set to work on genuine challenges, from industry for example. Now and again they are astounded to find themselves spending 70 percent of their time doing no technical work while completing a challenge, but doing other tasks instead: understanding stakeholders, understanding data, bringing everything together in a model, evaluating models built by other people. They are learning that they have to consider these kind of things before they can get a technical model to work. Just like in the real world.”

The Master of Data Science & Artificial Intelligence

The Master of DS&AI is intended for students who are interested in studying, researching and combining sophisticated data analysis techniques with AI methods and techniques. The main aim of the program is to train AI engineers capable of building reliable systems while bearing in mind the people who will use them. Fahland: “For the high-tech systems developed here in the region, for example, this is important; high demands are placed on their reliability.”

Students are taught a broad basis in the master's, with courses on the fundamentals of artificial intelligence, data visualization, ethics, algorithmics, data mining and management, and machine learning. The compulsory courses Ethics in Data Science & Artificial intelligence and the Data Intelligence Challenge connect fields of expertise while also providing a real-world context. Thus, in the latter challenge-based course the master's students are introduced to challenges from the field, working with stakeholders from industry for whom they must solve a problem.

From the start, Fahland held the view that ethics should be a compulsory part of the curriculum. Together with Full Professor of Philosophy & Ethics Vincent Müller, the team spent eighteen months working on this aspect. Repeatedly drawing on input from the master's lecturers in order to find the common denominator which they call “being a ‘good’ engineer”. “Some courses simply are technical; you shouldn't aim to weave ethics into everything, only into the right places. And I want it to mean more to students than a course they have to check off their list. They must look further than the engineering, consider its implications. We have to teach them not only how a good solution works, but also how things can go wrong. Errors are such a rich source of learning, and by giving examples you bring that alive. Ultimately, it makes them better, more rounded engineers.”

Feedback

For Fahland, feedback is vital, all through the development process, and now that the master's has been running for a quartile. From both colleagues and students. “I don't want to wait until our students graduate before making adjustments. We have them evaluate the courses while they are still in progress, we send them questionnaires and often receive detailed feedback. We will also be sitting down with our students to ask them whether they are aware of what they have learned, whether they have spotted any mismatches. We are being ably helped in this by study association Pattern,” says Fahland. “Our students are tremendously motivated, and feel a strong sense of engagement with their program. Together, they are making sure this master's is a success.”

“In this role we can discuss how things can be improved”

Matthijs Keep and Natasha van den Berg are student members of the program committee for the Master of DS&AI. They recognize Fahland's need for feedback from students. “Dirk is one of the few people at our university with whom I can be brutally honest; I know he won't take it personally,” says Keep. “We can always tell him what would work well for students and we know he'll give an idea serious consideration, however busy he is,” adds Van den Berg.

“With it being a new study program, I feel the need to voice my own opinion, to improve things. Members of the academic staff are also actively seeking our views; it's nice to be able to share them,” says Van den Berg. Keep endorses that: “I'm always looking at my own courses with a detached eye. That sets the brain cogs whirring: what's the workload like, do the courses fit together logically?”

Since finishing the Bachelor of Data Science, they have both been involved in education and in the setting up of the new master's. Both are experiencing at first hand that things don't always run as smoothly in practice as the theory would have you believe. “Any course can be harder or easier than average, and sometimes courses running concurrently make a less than ideal combination. The reality of this balance only comes to light when you are studying,” says Keep. “I really like that this role gives us the chance to discuss how things can be improved.”

Master's students DS&AI Matthijs Keep (right) and Natasha van den Berg.

Concern

For Van den Berg and Keep the huge intake is a cause for concern. “Most courses are big and cater for a massive audience. You realize that keeping the work pressure manageable for lecturers is an issue that's been looked at carefully. So, for example, they are more likely to set multiple choice questions in an exam because that's quicker to grade. It begs the question whether in the long run this is good for quality,” says Keep.

Van den Berg feels the personal attention lecturers are able to give can suffer due to the group size. “Everyone wants to ask questions, and even though there are five tutors on hand, it's a long wait until it's your turn. The high intake is a tricky issue because you don't want to turn away people when they are so keen to take this program.”

“But how are so many students going to graduate at the same time?” wonders Keep. Van den Berg: “At times I think, ‘that's still a long way off,’ but at the same time I know that everyone is going to want a final project they enjoy. Understandably, since it's nine months of your program.”

First cohort

All things considered, they are satisfied: “It is a new master's, but it is a well-organized new master's. Many students who start it don't even realize that they are among the first cohort,” says Van den Berg. They are happy that this program was available to them. Both took their bachelor's at JADS, a collaborative alliance between TU/e and Tilburg University, but were looking for a master's with a stronger emphasis on technology rather than entrepreneurship. Van den Berg: “A lot of students in my cohort felt the same way. In all likelihood I'd have gone to Amsterdam, but there's a stronger emphasis there on finance. I'm more interested in the focus on statistics offered by this master's.”

 

 

Bonus points thanks to AI system

Now, in the second quartile Fahland himself is teaching the course Advanced Process Mining, in which he reveals the fundamentals of process mining. The theories in data science and AI students learn in the master’s program he also applies in practice by making use of social reading software. He challenges students to highlight and discuss among themselves the texts they must prepare for the lecture. An AI system assesses the quality of their comments and hands out bonus points for active reading. “I don't have blind faith in the system, I only use it as an aid. But what I really like is that students are now asking me, how exactly does this AI system work. And then I analyze live in the lecture the event data that is attached to the comments they made and use process mining to generate a diagram showing how the whole class studied the text. They are seeing that what we are teaching them is being applied in practice, and that completes the circle.”

Challenging student numbers

As soon as it was announced that the master's was coming, they were inundated with student applications. In September 2021, 155 students started the program. “At a certain point, it looked like we would get 200 enrollments. That had shades of a worst case scenario, although I find that a terrible term in the context of launching a new master's. Even now students are still joining, for example, from the Bachelor of Data Science, who needed one more quartile to finish up their bachelor education. And we have premasters students who have to polish up their knowledge before being allowed to start the master's. I think towards the end of the year we'll be at 170 students.”

These numbers are an occasional source of mild concern for him and his fellow lecturers, but Fahland also knows: “At our department we are used to delivering master's education on a large scale. So in that respect we're okay, although we do need to keep an eye on our younger, somewhat less experienced lecturers. We have hired the new colleagues we needed, without exception ambitious researchers who are eager to teach students about their research theme. They are going to have to mature fast as lecturers, and we'll need to help them do that. The same applies to their supervision of students nearing graduation.”

"Our students are motivated and eager to learn. That makes teaching them so enjoyable." Photo: Loraine Bodewes

As a lecturer, I'd find it more enjoyable if I had more time for each student

Dirk Fahland, driving force Master of DS&AI

“Despite the challenges of lecturing to so many students, I do enjoy it. The students are motivated and eager to learn, and seeing that always reminds me why I'm doing this. I can just as easily hold riveting discussions with 130 students as with fewer, and it adds to the diversity. I'm sometimes asked questions that haven't occurred to me in my research role. But as a lecturer, I'd find it more enjoyable if I had more time for each of them and can build a much closer lecturer-student relationship. And that's exactly what you need in the second year of the master's, when they are working towards their graduation assignments.”

Right student on the right program

If student numbers are to be kept in check, it will be important to attract the right students to the program. “None of us wants to see a ceiling on student numbers or an intake freeze put on this program. Industry is champing at the bit for these graduates. We have an obligation to supply good engineers; society is calling for them. But you have to select students well by asking probing questions as to why they want to take this master's. And, for our part, we need to be very clear about what they can expect. So that we get the right students on the right programs.”

Fahland hopes that in the meantime more programs will be introduced in these fields, so that students have more choice when they are choosing the program best suited to them. “At our university people are currently working hard on a new interdepartmental Master of AI Engineering Systems. It is good to know that we'll soon be able to offer our students a broader choice of AI master's.”

Growth process

Fahland is now the program director of the master's that he nurtured. It has been a growth process for himself too. “I am a very different person than I was two and a half years ago. I was young and junior and had no idea what I was getting into. If I had known then how much time and effort this would cost, I'd have thought twice before diving in. There were times when my entire week was consumed by the master's and I was busy with it night and day. I think, with hindsight, that my wife would have advised me against doing it,” says Fahland with a wry smile.

The second quartile of the first year of the master's is underway and there's still work to be done. Aside from the high student numbers, the biggest challenges the program director foresees are filling the vacancies for academic staff and continuing to involve the present teaching staff. “I hope I can continue to infect them with my enthusiasm, but they are so busy teaching they can barely find half an hour to brainstorm with me, yet I know they are enthusiastic about AI and data science. Nonetheless, we are still making headway. Happily, I am blessed with inexhaustible optimism.”

From our strategy: about Talent

Talent, that is what it is all about at our university, at all levels. Of course, we are talking here about students, but we also mean professors, lecturers, researchers and support staff. We see it as our task to help them further develop their talents. Our students are the next generation, and we are equipping them to work on major societal challenges. In this, our lecturers play an important role.

Read more about our Strategy 2030.

Brigit Span
(Corporate Storyteller)

Keep following us