Hi, I'm Herman, years old, and I live in Groningen. I'm a technology enthusiast with a Master's Degree in Artificial Intelligence, a passion for creativity, and a website to tell you all about it.
In my spare time, you'll find me enjoying life with friends, family, or with my personal projects.
Unfortunately, I tend to come up with new projects faster than I can implement them.
One of my favourites is Style Transfer, where I transform my own photographs into works of art.
I've also built a Marktplaats Crawler to track prices of second-hand cameras & lenses.
Furthermore, I love to experiment with a Raspberry Pi: I've built and programmed a custom-designed wirelessly controllable robot with some motors, a battery pack, and LEGO.
By now, I hope I've piqued your interest. If you have any questions or compliments you'd like to share, or if you're looking to find out more about me as a photographer, you can use the buttons that I've conveniently placed below.
For the Master's degree, I passed all courses with an average grade of 7.6. My specialization of Computational Intelligence & Robotics means that more of a focus was put on the machine learning aspect of the degree programme.University of Groningen
I collaborated with a group of students enthusiastic to learn more about entrepreneurship in practice. The extracurricular course also featured guest lectures by successful entrepreneurs in the area.University of Groningen
In the Bachelor's degree programme, a large variety of courses were taught, such as programming in C, C++, Java & Python, Robotics, Linguistics, Psychology, Philosophy, and Biology.University of Groningen
I passed the Dutch VWO+ with exams in e.g. Physics, Chemistry, Biology, and Mathematics ('Wiskunde B').RSG Ter Apel
For my Master's Thesis in Artificial Intelligence, I researched automatic language identification of sung languages in music. I hand-crafted a language-labelled 6L5K Music Corpus, containing thousands of songs for each of six languages. For classification, I processed 3-second audio fragments from the corpus specifically containing vocals, computed mel spectrograms and MFCCs as features from this, and trained two state-of-the-art neural networks for pattern recognition in my thesis: “A Machine Learning approach to Automatic Language Identification of vocals in music”.University of Groningen
I've been a Teaching Assistant for the courses Introduction to Logic (1x) and Language & Speech Technology (3x) at the University of Groningen. This includes giving tutorials, grading homework, answering questions, and supervising exams.University of Groningen
Combining Artificial Intelligence with my love for photography, I trained a Multilayer Perceptron to reduce noise in images using additive white gaussian noise added to images from the CIFAR-10 dataset in my Bachelor thesis: “Improving Natural Image Denoising Using A Multilayer Perceptron”.University of Groningen
For a full weekend, we did research on a huge dataset of phone sales and metadata by Belsimpel. We tried to answer the question whether phone returns can be predicted, before a potential buyer makes the purchase. In their data, we could!Belsimpel
For this project, I research, apply, and improve state-of-the-art style transfer methods in order to turn my photographs into true works of art.Artwork by Herman
I built an automatic crawler using proxies to track prices of second-hand listings on Marktplaats. Whenever the crawler is run, new, unseen listings of the specified queries are appended to Pandas DataFrames for further analysis.Crawler on Github
Using a Raspberry Pi 3, DC motor, H-bridge, servo motor, spare battery pack, and some LEGO bricks & gears, I made a custom-built robot vehicle that can be wirelessly controlled via SSH (keyboard teleop) to drive forward, backward, and steer. The plan is to add a camera and ultrasonic sensor, and use reinforcement learning to train the robot to drive around without colliding.Quick snapshot
I developed and published a camera application for Android to the Google Play Store.
“If there's no software that does what you want, then it's either time to create it, or time to realize it's a terrible idea.”
– Herman Groenbroek (2019)
Desktop software that visualizes past events from digital files such as photos, videos, calendar events, created documents, location history, and other personal data you may have gathered, all in a single calendar view. It is intended to be the go-to application to check what you were doing on any day in history, given that you have data made or modified from that day. As of now, it is still in early development.Stay tuned for a showcase!
I love to implement neural networks such as CNNs, RNNs, and MLPs to solve complex classification and regression problems. However, I will also make sure that the solution fits the complexity of the problem in the first place.
In Python: TensorFlow, Keras, PyTorch, Numpy, OpenCV, SciPy, Scikit-Learn
If you hand me buckets of data and some time, I will happily dive in and find correlations in the data, interesting outliers, predictors, and visualize all of this in graphs that are informative and pleasing to the eye.
In Python: JupyterLab, Pandas, Matplotlib, Seaborn, Altair, SpaCy, Essentia
I taught myself Android app development, GUI design for cross-platform software, and web development for the website you are currently looking at. I love to program in Python, and I also do well in e.g. Java and C++.
Although I studied Artificial Intelligence, my hobbies and interests lie well beyond this scope. Within IT, I not only develop deep neural networks, but I also like to build robots, apps, and desktop software. Outside of IT, I love all fields of science, photo- and videography, playing instruments, and even painting.
What will you teach me?
A lot of enjoyment comes from creative expression. I love to take pictures that make the viewer think. Or create music for harmony. In business, my creativity translates to taking an additional step to think ahead. “What are the possible ways to approach this problem? And what would be desired, but has not yet been requested?”
No matter what I do, I like to do it well. When I build something, I make sure that all requirements are met, and I will especially consider the less common use cases where things might go wrong. As such, I have an inherent drive towards flawlessness.