“I was interested in studying abroad in Germany because I love the language and I know there are lots of great computer science opportunities in Germany. And then I heard about the UAS7 research internship programme (IP)* through the study abroad office at my university. The IP programme offers research opportunities in different fields at seven different German universities. It was an obvious choice for me and it was extra-enticing because I really enjoy hands-on research and I know that internship experience is very important for job opportunities. So, I was thrilled at the opportunity to study abroad and do a hands-on research internship at the same time. During my computer science studies at Binghamton, I have participated in various machine-learning research groups in the fields of bioinformatics, object detection and bio-acoustic monitoring. At HAW Hamburg the MARS* research team offers IP research projects with an insight into multi-agent modeling and simulations. I was particularly drawn to this research topic because I am fascinated with thinking about cities as a simulation and modelling their populations in order to produce real and accurate advice for urban planners and engineers. I was especially interested in discovering how machine-learning, specifically reinforcement learning, could be used to make more human-like agents. I also felt that Hamburg was a city that I could envision myself living in and enjoying very much.
After I was accepted to the UAS7 internship programme, I spoke with the head of the MARS group, Professor Thomas Clemen, to learn more about his work and to discuss what specifically I would do there. We talked about my skills and interests and settled on a way for me to contribute to their work that was both interesting for me and important for them. In May 2020 I was all signed up, but the developments in the corona pandemic made it clear that travel to Germany in September 2020 would be impossible. I initially thought I would have to give up my plans, but the more I thought about it, the more I knew that the opportunity was still a great one and that I should not pass on it. I knew I would learn a lot from the research and that I would be able to contribute in a meaningful way even from America. So, I contacted Professor Clemen and we discussed the option of doing the research remotely. He was very supportive of the idea, and so I talked to my academic advisors and together we were able to come up with a plan, in which the Hamburg research would fulfil a 4-credit computer science class, the final four credits needed to complete my degree.
The remote internship lasted from 19 October 2020 to 11 January 2021 and my experience as a research intern with the MARS Group at HAW Hamburg was challenging, educational and fun. In my first meeting with Professor Clemen we talked more specifically about the timeline of the programme and how I would proceed with the research. It was a slightly overwhelming at first, because I was given a large (but very concise and well-written) codebase in C#, a programming language I had not yet worked with. My task was to implement an extension to it, so that the simulation agents (software robots) could make decisions about how to travel in a more human-like manner, leading to more accurate simulations. I was planning on using a branch of machine-learning called reinforcement learning. The basic framework for the internship was a one-hour meeting with Professor Clemen at 9:30am EST every Monday. Between meetings I worked on implementing what we had discussed. I was also added to their Slack group, so I could message Daniel Glake and Florian Ocker, PhD students in the MARS group, who were very knowledgeable and helpful. I contacted them when I felt I had encountered a problem that I really could not figure out myself, but also saved questions for my next weekly meeting with Professor Clemen. Around the third week, I started understanding the code a lot better and developed a clearer grasp of the problem that I was looking to solve. Needless to say, this took hard work every week but paid off immensely. Every week I would prepare several slides and notes as a presentation of my work and of ideas going forward, which made the most of the meeting time and kept the project moving forward. Every Monday I would leave the meeting with a renewed excitement to implement the ideas that Professor Clemen had given me and the ideas that we brainstormed together.
Working remotely has its challenges, but Professor Clemen was very reliable and I had his support throughout the internship. The time difference was a bigger problem in the beginning, as the PhD students in Hamburg would rarely be able to get back to me the same day. But as I became more independent, that stopped being an issue. I had a lot of freedom regarding which research avenue to take and how I went about the research within that avenue. Professor Clemen was very knowledgeable and had a clear idea of what we were generally looking to accomplish. His guidance was extremely valuable. I feel that the research I did has the potential for a publication, potentially to be very useful for the MARS group and best of all, it was extremely interesting and fun for me! I would recommend this experience highly to anyone considering the UAS7 IP programme with HAW Hamburg – either remotely or in Hamburg, once travel is possible again. I have learned a great deal about how multi-agent modelling and simulations are created. I have also learned a lot about reinforcement learning – the theoretical aspects as well as the technical. Another reason for sticking to the plan of doing the research internship, was an idea I had of what I wanted to do after graduation. Having learned more about the HAW Hamburg, my current plan is to move to Germany and pursue my Master of Science degree in computer science there, so I am working hard to get my German up to the required level. I already have great connections! I am also considering pursuing a PhD afterwards.”
If you would like to learn more about this experience, Gabriel Steinberg is happy to answer your questions: gsteinb1 (@) binghamton.edu