Bastiaan Grisel – Media and Knowledge Engineering

The experiences of a Media and Knowledge Engineering Student at the TU Delft.

Back in Business

Holidays. Finally a few weeks to unplug and do absolutely nothing. So I thought. In the first week of the holidays, after finally handing in all (except one) of my reports I went to Brighton and London to present out VR data visualisation. After that, I decided to get some new furniture for my room, get some new paint on the walls since I hadn’t had a chance to do so after I moved in April. A week of sailing the IJsselmeer followed and after that I finalised my Machine Learning report. As if that wasn’t enough, I performed with my band at the freshmen’s weekend of my study association which was super-cool. More details below.

Let’s first dive into my experiences in Brighton and London. Bert and I had been working like crazy to get our VR demo up and running for the conference we were attending 15-16 July in Brighton. Everything went silky smooth and when after setting up on the 14th and making some last-minute improvements we were ready for people to try our demo. And boy they did. I started talking when the expo opened only to stop when the expo closed. Everyone wanted to see what we’ve done and people were incredibly excited to see the data in VR. It was so cool to show everyone what we’d built but at the same time incredibly exhausting to tell the same story over and over again.

In the end, we didn’t win the $20.000 price but the $5.000 we were given at the beginning of the contest covered our expenses anyway (including beer and fancy dinners) and losing to professionals in the gaming industry was no reason at all to be disappointed. The whole experience has been great and shows that you should definitely go and participate in these kinds of events. Just try applying to some, if you do not get in you just continue your usual business and if you do you’ll get an amazing experience in return. Even though the Big Data VR challenge is over, there is much more in the pipeline. Our research team would love to continue working on our solution and we are looking for way of funding this, hoping to hear something of that in the next month. Next to that, I have some very cool opportunities for my thesis project via someone I talked to in Brighton. I hope that will work out but it is all incredibly exciting.

Watch the final video of the Big Data VR Challenge below:

After spending a few days in London, we returned to Delft feeling proud, content and honoured to have been part of this experience. The following week I spend relaxing at my parent’s home until I decided that I should do something with my room. After a few days of painting and visiting large interior shops my room is once again a cool and cozy place.

That week, I also intended to finish my Machine Learning report which was due somewhere in August. With all the room-decorating business this did not go according to plan. I did, however manage to get first place in the competition that was hosted for the Machine Learning course. The solutions to the project needed to be uploaded to Kaggle (a site that hosts Machine Learning competitions) and you were ranked according to the score of your solution. The assignment was a classification problem and the task was to assign labels 1 or 2 to a set of unknown objects. I did some testing and discovered (by accident) that the first few objects had label 1 and the rest of the objects label 2. I submitted a file with the first 4.000 objects being class 1 and the rest class 2 and my score was 100% correct. This wasn’t of course entirely fair play, but it did shake up the scoreboard a bit since the best attempt had been around 80% correct labelling. Soon after my attempt, a few others noticed the same thing and also gained a 100% score. Since I was the first one to discover it I kept the first place (muhahaha).

The next week I went sailing on the IJsselmeer with my dad. My dad had purchased a boat a few years ago and I have been using it extensively throughout secondary school and the first years of university. We did a complete refit of the boat, redoing all of the paint jobs, replacing all nuts and bolts, really fixing every part of the boat. After putting in so much work it is great to take the boat out to go sailing. Unfortunately, due to all my side-projects, I did not have the time to go sailing for a weekend very often so I was happy to be able to go away for a week.

Luckily I managed to finish my Machine Learning report after return from sailing. Unfortunately I do not have pictures yet, but I performed with my band for all Mathematics and Computer Science freshmen. The crowd loved what we were doing which made it a very rewarding experience.

Today, I just enrolled for some new courses I’ll be following in september. Security and Cryptography, Advanced Digital Image Processing, Seminar Internet of Things and Embedded Real-Time Systems. Almost all courses are built around a project, we are going to build a drone from scratch in the Embedded Real-Time Systems course, and will be building some smart connected device in the Internet of Things course. I’m really looking forward to the coming quarter but it will be once again incredibly busy with 19 ECTS. We’ll see how it goes!



Note to self

Man, I can’t believe it; the academic year has almost come to an end and I’ll have officially finished my first year of my Masters programme. After five astonishing years at the Delft University of Technology I’ve got only one left, which feels a little frightening to be honest. Firstly, because I’ve only got one year to do whatever I’ve always wanted to do in my academic career and secondly, I have to formulate my plan for taking over the world once I’ve graduated. Luckily, I lived my student live to the max over the past years so I haven’t many things left on my academic bucket list. The figuring out what to do after my studies is quite hard though.

I mean, the world is at our feet and it feels so exciting. Especially as a Computer Science student you’re able to work practically anywhere and if there is an industry that knows no geographical boundaries it is IT. If I want to, I could travel the world with just my laptop and do small programming jobs in every country that I visit to pay for the rest of my journey. A great way to see more of this amazing planet, experience new cultures and learning a thing or two about earning your own money. If I want to, I could start working at large consultancy firms that operate globally. I could advise the top management of the largest companies on the planet on what strategy to employ to maximize revenue or penetrate new markets. I would travel the world, work in teams with the brightest minds and earn more money that I would be able to spend. If I want to, I could use my Computer Science knowledge to tackle the world’s most difficult problems at innovative companies like ASML, Philips, NASA, CERN, you name it. If I want to, I could work in software engineering and make an impact with delivering large-scale systems for healthcare, taxes or enterprise resource management. Heck, I could even start my own company! If only I knew what I wanted.

If only I knew what I wanted. “What do I want?” I’ll ask myself, without even knowing if there is something I would want or if it is even necessity to actually want something. And what is this something? What does it mean? What I want to achieve in life? What I want to earn in terms of money? What I want my day-to-day activities to be? What I want to contribute to society? What I want to be remembered for? It seems that to answer the question “What do I want?”, I need to define both the ‘what’ and the ‘want’.

The ‘what’ part is the simplest one and the part I’ve put a lot of effort in over the last years. I visited many companies to see what they were doing, I was a board member to discover what industries there are and what opportunities there were in the academic world, I have talked to many people about their career paths and the choices they have made in their lives. I have visited entrepreneurs that have their own business, I have had cup of coffee with people from the creative industry. I feel that I have seen a lot of ‘what’. The ‘what’ involves observing a lot, absorbing, listening. As long as you take a bit of initiative, the ‘what’ is quite easy to figure out. If you do not have a view of the ‘what’, then just go out there. Discover what ‘what’s there are. The what is waiting out there to be explored. The hardest part is yet to come.

It is the ‘want’ part. It is the part in which you need to digg into your own ‘I’ and search for… for what? Many people I’ve talked to said that people like myself need to search for their passion in life. What it is that I truly wantdesire, get out of bed for every day. The thing that makes me happy, gives me joy. Much like a musician could have a passion for music and maybe conveying emotions, a physicist could have a passion for the measurable world, a mathematician could have a passion for the abstract world, a sportsman could have a passion for cycling, a singer has for singing and so on. What is my passion? I wouldn’t know the answer to that question. Do I even have one single passion? Do I even have any specific passion at all? Is it necessary to have one? Is having a passion a requirement for living a fulfilled life? Is living a fulfilled life something I would above all want? I. DON’T. KNOW.

Who am I asking these questions to? They almost seem to be directed to some omniscient narrator who I expect to provide me with a logical, sensible answer. If assume this to be true, I think I’ll have to wait quite some time to hear the answers. It seems rather pointless to direct these questions to some higher power. And for a reason. These are not questions I will get the answers to by anyone other than myself.

The emphasis in the sentence “What do I want?” is not on the ‘what’. The ‘what’ is merely an observation. A collection of options you can choose from. The emphasis is not on the ‘want’. Most definitions of what there is to ‘want’ are formed by a collective opinion of people describing ‘passions’ and ‘purpose’ implicitly imposing that these two things even exist. The emphasis in the sentence “What do I want?” is on the ‘I’.

There are no predefined answers. I don’t even think that there is a single way of coming to an answer. This is what makes this question so damn difficult. One might even wonder if it is even worth posing this question at all? In my opinion the answer is a big fat YES. I don’t know if I will ever arrive at an answer and how long it could possibly take me, but I do know that the mere journey one embarks on when posing this question is one worth taking.

Skip the Intro

So many interesting things have been happening over the past few weeks, it’s just been a rollercoaster with projects. I’ll share an update on each one.

Big Data VR Challenge

During the second workshop at the 29th of May, we have seen what the other teams have been doing and how they are getting on with developing their -mind the buzz words-virtual reality solution for big data visualisation. This time, we did not have to fly to London, but a virtual meeting was arranged. I’ve never been in a 6 hour virtual meeting before, but it is quite fun to talk to people from Australia, Canada and the UK from your own home.

YouTube Preview Image

Just a small recap: the data that we have to work with consists of 30.000 cases in which people, who lived in the years 1550-1610, visited two general practitioners who documented all of these cases in large books, called the Casebooks. These cases have now been transcribed to a machine-readable XML format.

The initial concept we came up with for this challenge involved you sitting in the enter of a 50m-wide spherical dome in which individual data-points (cases) would be flying around. When we developed this concept and took a look at the 30.000 data-points we had visualised, we quickly realised this was not the right way forward. The sheer size of the data was impressive, but the vast amount of data points made it impossible to discover any interesting insight. Grouping, scaling and linking the data might help, but we did not think we had the horsepower to recreate a virtual reality equivalent of MatLab in this three-month challenge. And although the whole experience was fun for the first ten seconds where you literally stare in awe to all the spheres floating around, there wasn’t much to the whole experience that motivated you to dive into the data. Moreover, these abstract representations of the inherently personal data which we had to work did not seem to use this data to its full potential.

From the different prototypes we have built, we learned a few things: this data is about the people living in that era and the relationships between them, the final product needs to encourage the user to explore this data and let them find interesting relationships. With these two elements combined, we quickly arrived at some sort of Quest-based game in which the user gets rewarded for the number of relationships he or she finds between different people in the dataset. Instead of aggregating all of the data, we would like users to explore the relationships between individual people in the dataset.

This is a screenshot of the demo we have shown. The yellow figures are the people and the goal is to build a bridge from one end of the river to the other end.


Computer Vision

The goal of my project for Computer Vision is to develop a vision-based system that is able of counting Euro coins faster than a human. The final application that I envision can be installed on your Android phone and works as follows: you throw some coins on a table-top and make sure they do not overlap, you take a picture with the app, the app asks you to classify one coin and the monetary value of all the coins is shown.

The first prototype will be built with MatLab, since many libraries for analysing images integrate smoothly with it. Once a working solution is found, I will try to implement it in Java and make it run on an android phone (or maybe some web-appish thing).


The current idea is as follows. First, the coins need to be identified in the image. Since coins are circles when seen from the top, I can search for circles in the image and take them as the coins. Note that Euro coins are uniquely determined by their color and their size. So if I can estimate these two properties in a good way, I can determine the value of a coin. This proved to be a bit difficult since the size could not be determined in an accurate way and the color is dependent on the lighting. Ouch. The next thing I will do is try to look at the actual shapes on the Euro coins, the downside of that is that all of the backsides to the Dutch Euro coins are the same. As you can see, there is no free lunch in Computer Science. We’ll see how it ends.

YEAH! We’re making a VR Game!

I promised to tell you about my experience with the Oculus Rift DK2 and the Kinect v2, so I’ll do that. But there is more news; I got selected to participate in the Unreal Engine Big Data VR Challenge in which six teams are going to make a data-driven VR game. This is gonna be big.

A few months ago, the study association of Mathematics and Computer Science at the TU Delft bought an Oculus Rift, a graphics card that could keep up with it and a Kinect for Windows v2. The idea was that students could just play around with these gadgets and do cool stuff. And so I did. If you have tried VR, you know that it is very difficult to describe the experience and that is really a matter of seeing is believing. So if you want to experience what VR looks like, just pass by the study association.

After playing countless of cool demos and wandering around beautiful VR worlds (I could stare with awe at the simplest things like a wall or a chair), I wanted to make something myself. I installed the Unreal Engine 4 (UE4), which Epic just made freely available and followed some tutorials on how to make a simply VR demo. It was incredibly satisfying to actually create something again after a few theoretical courses at the uni. Since there was also a Kinect v2, I used the K4U plugin to be able to see myself in VR.

YouTube Preview Image

As I was browsing the Unreal site some time later, I stumbled upon a blog post stating that Epic Games an the Wellcome Trust were hosting a challenge in which six teams were to make a data-driven VR game. This was exactly what I was looking for. I’ve been studying machine learning for the past year in my masters and over the last weeks had been diving into VR. Although I had made some games in the past (during my bachelors) and am currently taking a course in 3D Computer Graphics and Animation, I did need to find someone who had a bit more experience in game development. Luckily, there are tons of these people at the uni and I knew that one guy I worked with on some project was developing his own game engine. He was incredibly enthusiastic to participate and we decided to sign up. We figured we might need some more people on our team (I have done some Graphic Design in the past, but no real 3D modelling of some sort) but since the application deadline was approaching, we applied as a team of two and let the organisation know that there might be more people involved.

Two days later I got a reply that we were selected! We had to attend a kick-off meeting in London in a week so we booked our flights and hostel the same day and had a great experience the week after. Upon arriving home, we started brainstorming and we have just started to make a prototype around this time. We’re now thinking of building X-men’s Cerebro, I’ll keep you updated if I have any cool things to show.


For a course called the ‘Business Development Lab’ we are investigating the use of hyperspectral imaging for the analysis of plant health. The course is highly practical and we visited multiple companies over the last few months. The focus is on the business model behind the idea and making sure that the problem you are solving provides actual value for the company. We started out with a case about DNA analysis for humans, then focussed on DNA analysis in plants but then discovered that this technique was too cumbersome and that image analysis could provide enough information to tackle the problems that plant growers have. After visiting with Sion Orchids and seeing how orchids are being grown, we have seen that they have to throw away many plants each year due to diseases that are detected in a late stadium. If we could detect anomalies in an early stage, this would lead to a huge cost reduction for them. In the following months, we need to elaborate on this project and create a solid business plan.


All in all, this period will again be a busy one with many exciting projects coming up. I’m doing a computer vision course for which I need to make a project, a computer graphics course for which I need to design a computer game, a machine learning course that has a final assignment, the business development lab and the VR game. If I’ve got any updates, you’ll see them here.

Take care!

Big Data Workshop

It’s been a few weeks and a ton has happened in the past month. I’ve attended a workshop-week, played with the Oculus Rift DK2, Kinect v2, spoke with people about DNA-analysis, organised a programming championship for freshmen students and moved to a new (larger) room in my student house.

Let start with the workshop that I attended. Ten students and a few company employees were selected to take part in this workshop about Big Data. I always get a bit anxious when hearing the term Big Data, assuming it will be yet another event organised by people who do not really know what they are talking about. This was not true at all for this event. It was organised by two students who participated in the yearly ThinkTank, a group that tries tackle a global problem in a period of three months (which is more sensible than it might sound), and I found that if these students were passionate about this Big Data problem, the workshop would be a success. And it was. I was expected to be present at the SAS institute in Huizen, which is this gigantic mini-castle surrounded by trees and wildlife: beautiful.

SAS Institute

The building was extraordinary, beautiful old rooms with decorated ceilings; you would almost forget that it was an office building. We had a cup of coffee and moved to a smaller building (the coach house) where we would spend most of the week. We introduced ourselves and the remainder of the day was mostly about visualising data and presentation techniques.

The rest of the days were organised by different companies. On Tuesday, GoDataDriven (GDD) gave a workshop about using R and Python for data analysis. It seems that especially R used in almost all companies doing something with data analysis while I thought of it as the iffy open-source software that I was forced to use during my statistics class. Entirely not true! It was fun to hear their opinions on data science and big data and how companies were trying to become more data driven (and where they failed/succeeded). A very interesting day and it was great that these people sacrifice a whole working day in order to tell us about their findings in the field. I’ve made a ton of notes with such insightful ideas that I will undoubtably share with you some time.

There was also one guy from GDD who told us that he toured the world with his laptop, so he could work on some programming projects on the road to finance his journey; a brilliant idea. He could go anywhere, as long as there was wifi available (which is pretty much everywhere, since all tourists want to upload their selfies). Of course, he encouraged us all to go and maybe, if I haven’t gone abroad when I graduate, I will do something similar.

Presentation Teradata

During the rest of the week, we had talks from the Dutch Railroad, Eneco, Teradata and SAS. It was super cool to hear what issues were at play at these companies and it provided me with some well-needed context for all the hard work at the university.

There were two more things that were amazing about this workshop, and one was the food. We had breakfast, lunch and dinner at the SAS Institute and it was like dining in a proper restaurant. We had entrecote and fillet of sole for lunch accompanied with freshly made soups and freshly squeezed fruit and vegetable juices. As much as I looked forward to the next workshop day, I was excited to taste what the in-house chef had made this day. He was also incredibly friendly and explained passionately about the food he was producing.


The other thing that really stood out during the week was a phenomenon called: the energizer. You all know it, when having an intensive day with lots of talks and hands-on sessions you feel really exhausted somewhere around 15:00h. Not that the talks are not fun anymore, but the concentration has vanished and you just need some new energy. The organisers of the workshop had prepared some so-called energizers. It often involved the whole group going outside (the weather was great!) and doing some hilarious group game.

In one game, everyone got a piece of paper on which an animal name was written. There were two pieces of paper for every animal. Everyone then needed to close their eyes and make the noise of that animal and locate their counterpart. It was hilarious, some people didn’t know how a particular animal sounded and just gave it their best shot while some passionately trumpeted like elephants and others quacked like a duck. Great that such a group of university students is able to step out of their comfort zone collectively.

Another great one involved playing rock-paper-scissors against one another. If someone lost, than he became a supporter for the winner. This accumulated and resulted in the group being divided into two parts, each chanting for their winner.


A last one which was really inspiring was a game which you could easily do wrong (I’ll spare you the details), but every time someone made a mistake, instead of laughing at them, everyone needed to applaud and chant to this person and pretent as if this person has done something great. I noticed that this gave such a great amount of energy, by both cheering at people as having people cheer at you and it is a great way to illustrate how I think about celebrate failure.

After having such a laugh, you feel re-energized and ready for the next session. A great idea which you should totally try in your next group session. Inspiration: here and here .

Whoa, this blog has gotten a bit out of hand but I did not want to spare you this experience. I have a lot more to write about but I will save it for the next blog 😉 Have a good one!

Free time!

After a great week of riding my new awesome snowboard in fresh powder and having great weather we’re back in business with a fresh quarter. My schedule is quite relaxed with lectures on Tuesday and Wednesday. The remainder of the week needs to be spent on revising the lecture material, reading papers and making practicals. Luckily, this does not take up all week so I’ve got some free time to relax and work on some side projects.

One thing I want to show you from the previous quarter is the finished project of Data Visualisation ( The idea was to make a packet capturing application that would show where packets would end up. We did come up with quite a neat visualisation that shows exactly to what countries your information gets sent to. It’s really quite cool that this visualisation is working since packet capturing is a very low-level business and the visualisation is very high-level.
Datavis screenshot

I’ve just attended a lecture on 3D Computer Graphics and Animation, which I feared would be very mathematical and not much about the fun side of virtual worlds but this seems not to be the case. The professor prepares his lectures very well and explains everything in a clear and concise way. Not at all a problem to get out of bed for (the lecture starts at 8:45). We have just discussed how images are generated on a GPU and how 3D objects are mapped to a 2D screen via a process called rasterisation. Seriously cool stuff. The practical assignments of this course will involve some low-level programming of basic graphics (triangles) and showing them on the screen. It does not seem that interesting at first, but it’s the beginning of every computer generated image. At the end of the course, we should implement a game of some sort. Don’t know the details yet, but undoubtedly Elmar will find something cool.

Yesterday evening, I have finished a practical for the course about Machine Learning. The professors that lecture the course are from the Pattern Recognition field which I find very interesting since it focusses on real-life problems (based on data) but has a very solid theoretical foundation. This combination is very exciting since complicated formulas can be put to the test immediately. This also has the effect that the practicals are very theoretical and the one I had to make was quite difficult. It was about identifying correlations in data by means of least squares approximation. The practical showed that this method is very unstable and produces different results when the data used varies only by a few observations. We implemented a version of this algorithm that copes with this problem. Quite interesting, but very mathematical. I’m looking forward to hearing more about learning theory and what it means to learn a computer something. Maybe this week.

In my quest to get selected for some extra-curricular activity, I’ve tried: auditioning for a play and applying for a management consulting business course. Unfortunately, I did not get selected for either but two weeks ago I spotted a Big Data Workshop week in which 10 students are selected to do small data-related projects at cool companies like Teradata, SAS and GoDataDriven. The good news is that I got selected for this week which is incredibly cool. I’m super excited to see how data is used by these large companies and if the thing I’m learning at the uni also apply in practice. I will definitely keep you posted if I know more about this. I can recommend everyone to just apply for some things you kind of like and if you get selected, it will be a cool experience and if you don’t then you just continue with what you were anyways doing.

Speaking of extra-curricular activities, we managed to land a gig with our student band. We’ve played for two hours at a student ball and it was unbelievably cool Everyone loved it. It is incredibly satisfying seeing all of the practising coming together on stage. We’re now looking for some new gigs and have a few potential leads. We’ll see where this goes but we’re definitely not done yet!

This afternoon, I’ll be programming some cool application for the new Philips Hue lamps that the study association acquired. I’ll also write a manual for others so I’m not the only one who is able to program these new gadgets. I hope to get this done today so I can proceed with setting up the oculus rift at the end of this week! Exciting!

Exams are over! Ready for Q3

Just as it was busy before the winter holidays, the exams I’ve had in de previous few weeks kept me entertained for quite some hours as well. Luckily I had ‘only’ three, which ment I could spend enough time studying for each of them and get a nice grade (in theory). At the time of writing, I’m busy finishing two practicals which need to be handed in the next few weeks.

One practical is about data visualisation and in groups of two you have to come up with a cool visualisation of a certain dataset. Since I’d already been experimenting with packet capturing in Node.js, I figured it would be cool to turn it into some sort of easy-to-interpret visualisation so that laymen can develop an intuitive notion of data going in and out of their computer. We’ll be working on that in the following weeks and I will show you the result once it’s done. I hope it will look something like this:


The other practical involves building a computer program that can negotiate arbitrary issues with other computer programs. These issues typically involve basic scenarios in which there are a few parties that need to come to an agreement, for example, a few friends that want to organise a party together and need to decide in the food, drinks, location, music etcetera and each have different preferences. It is quite cool to think about this stuff, but the fact that the problems the software can deal with are simple and far from the truth that is a bit demotivating.

Luckily, I’ve got a week of snowboarding coming up where I’ll be going to the French Alps to ride my newly bought Arbor A-frame snowboard. I’m super excited to take it for a ride, it is supposed to be super fast, super stiff and awesome in powder. It going to be snowing quite a bit next week so we’ll have plenty of fresh snow! I’ve also heard that our apartments are right across the street from the après-ski bar, so it’s bound to be a tough week!


When I’m back in Holland again a fresh quarter will start. I’ve chosen to do the follow-up course of Patter Recognition, which is called Machine Learning. This course will dive even further into the techniques required to make computers ‘intelligent’. Super cool. There is also this course about 3D Computer Graphics and Animation, which is taught by a great professor called Elmar Eisenmann and will focus on the techniques required to make video games and animated movies like Avatar. I’m sure it’s going to be a lot of math, but if it serves a purpose it think it’s worth the hassle.

Another thing I’ve wanted to do for a while is expand my knowledge of innovation processes and starting a business. The EIT ICT Labs master organises a course called the Business Development Lab, which covers both of these subjects in a full semester. It will require two days a week of effort and I hope it is worth it.

Furthermore, I hope on getting my hands on some new gadgets in the coming months. If everything goes as planned, I’ll be able to play with the Oculus Rift (Virtual Reality headset) and hopefully develop some apps/games with it. I’m thinking of hooking it up to Kinect v2 sensor which is able to track 3D movements so we are able to become someone else in virtual reality. Hopefully I’ve got some time for it, but I’ll keep you posted if I’ve made some progress.

I also got an email a few days ago stating that I wasn’t invited to join in the theatre play I auditioned for, which is a bummer since I was quite eager to do some acting. Maybe it is for the better considering the other activities and project I’ve got planned for the next few months.

Anyways, I’m going to finish packing my bag and I’ll be back in a few weeks with another update!

2015: New challenges

Although it’s been a while since you’ve see me post something, this is not at all indicative of my activity outside of the blogosphere over the past weeks; quite the contrary is true actually. The end of the year is always quite busy with many deadlines before the christmas break (hope you’ve had a good one!) so I’ve spend most of december finalising some cool projects.

For a course on Pattern Recognition, we had to design a system that you could feed images of handwritten digits. Based on these digits, the systems needs to detect new handwritten digits it had not seen before. The theory behind this is extremely interesting and corresponds quite well with how humans learn from examples. The basic idea behind this so called classification is quite simple and in the image below you can find a toy example of it.

In this example, we have measured the weight and width of a few apples and pears and plotted the values of the apples as red stars and the pears’ values as blue plusses. We can now see that you can draw a line between the stars and plusses so that all stars are on one side and all plusses are on the other side. This line is called a classification boundary, since it divides the objects into distinct classes of stars and plusses respectively. Finding these boundaries in a somewhat intelligent way is one of the main concerns of the pattern recognition course! Now, if we want to know if a new object (the black dot in the lower right corner) is an apple or a pear, we just measure its weight and width and see on which side of the boundary the corresponding point would end up. In the case of the black dot, it is clearly an apple. This is the bare beginning of Pattern Recognition, since there are many challenges to consider when building more advanced systems, for example: what if the classes overlap so there is no line anymore that perfectly separates the classes? What if an object cannot be represented by two characteristics but only by 100? How would a classification boundary look like then? What if there are only a few objects available for which there are known to be from a certain class?

Classification problem

Another course I took was Data Visualisation. Since I’m also into graphic design, I really looked forward to this course. And I must say, it did not disappoint. The basic idea is as follows: you get presented various techniques and best practices of visualising all sorts of data during the lectures, followed by a practical in which you get an enormous amount of data which you should visualise. We have worked with gigabytes of game data and made analyses of the locations in which many players die (see below) and what different tactical movements highly skilled players have compared to novices.

Dota 2 hero kills

I am currently working on the visualisation of medical data from CT scans and the like. This data is different from other data since it is basically a collection of slices through an object which needs to be stitched together in an intelligent way to form a 3D model. Datasets for these models are freely available on . It’s really cool (and kinda weird) to experiment with data that originates directly from living creatures.

Furthermore I now follow a course on Artificial Intelligence, which sounds reaally cool and has a lot of potential but unfortunately is not of much interest. I would expect a course which answers questions like: what is intelligence? What are different forms of learning? How can autonomous systems work together to achieve a common goal? What is Natural Language Processing about? This list could go on and on but most of these questions get touched on in the most superficial way. I will try not to write away my frustrations but I get so upset by the staggering difference is in potential of this course and how it is given now. Since everyone deserves a second chance, I plan on following a course called Neural Networks, which supposedly is not really about neural networks but more about swarm intelligence which I find incredibly interesting and is given by the same department as the Artificial Intelligence course. We’ll see how it turns out!

Swarm intelligence

Of course there are many things happening next to my study. If everything goes as planned, my band will be performing at a student party at the end of February, which is really cool. We’re working on a setlist and practising every odd Tuesday to get everything together. Playing some music is also great to get your mind off all the daily routines and just relax for a bit. I play the keys btw.

Next to that I recently auditioned for a part in the theatre production called ‘Getekend’, which is about the lives of students during the second World War. I’ve always wanted to be part of a theatre production but most of the work I’ve done with theatre productions is backstage as a lighting and/or sound technician (also really cool). I had to study both a monologue and a dialogue, the former of which was a dramatic piece about jealousness and the latter a more informal breakfast conversation. It all went great and I’ll hear if I’ve got a part on the 20th of January. Exciting!

In the next few weeks I’ll be studying for my exams of which I have only three, which is manageable. After the exam weeks, I’ll be heading off to France to go skiing with around 50 people of the study association so that is something to look forward to. I’ll try to post an update before the end of the exam period and let you know about my exams and audition results! See ya!


Welcome to this blog about my experience with studying at the TU Delft and following the master track Media and Knowledge Engineering. In this first blog I will shortly (I hope) introduce myself and try to give you a summary of my experiences in Delft until now.

Me, me, me!
My career in Delft began in 2010 when I left the cozy town of Zeist to go studying Computer Science. Being a rather creative person, Computer Science did not seem like an obvious choice to many people for me to go studying. The truth is that many people view Computer Science as a dull, boring and inherently not creative study but luckily, as a student, it is not difficult to prove them -and yourself, for that matter- otherwise.

Especially at the TU Delft, the education is oriented towards group-work and learning by doing. This approach motivated me a lot, since I don’t mind solving complex integrals or learning about new pattern recognition techniques, as long I can apply this knowledge in a real-world situation. I strongly believe in the saying: “It is easier to invent the future than to predict it” and here at the TU Delft you are given many possibilities to go experimenting and (at least try) to create the future.

When studying pattern recognition techniques, one also has to create and implement the algorithms behind facial recognition systems and test them on real-life data (or use it to index my own photos). After a course in multi agent systems, a 4-month project is set up where each team needs to program a group of software agents to achieve a certain goal. This list goes on and on and it are these projects that bring to life the education given at the TU Delft.

As I said, I’m quite a creative person. I like to write, read, draw, think and come up with crazy ideas. The power of Computer Science is that I can implement these ideas by opening my laptop and typing away. This would be incredibly hard to do in say, Civil Engineering or Areospace Egnineering, but being able to consolidate my ideas quickly is and actually being able to interact with the result gives an enormous satisfaction.

Media and Knowledge Engineering
Although there are quite a few projects in the Bachelor phase of the Computer Science curriculum, most of the courses serve as preliminary knowledge for the real deal: the Masters program. During the first weeks of the program you realize that there are tons of possibilities for specialization within the realm of Computer Science and in every specialization there are enthusiastic people who love what they are doing. This makes it incredibly difficult to choose a direction in which to specialize.

I was really inspired by the works of creative people with in depth Computer Science knowledge, creating beautiful visualizations of complex problems that enabled every person looking at it to immediately develop an intuitive notion of the problem that was presented. Audio and video analysis also drew my attention and I was excited to learn about techniques that the people at, for example, Shazam employed. The data-analytics topic is quite a hype at the moment and I was a bit skeptical about it at first just because of that but as I experience it now it quite lives up to what people predict. In the first weeks I found myself analyzing large datasets and drawing relevant conclusions.

In the next blog, I will discuss some of the courses I have taken and the projects that I’m currently working on should take on a more concrete form. This will be around December.

© 2011 TU Delft