[Humans of FDL] Jakub Fil
Meet EO Researcher Jakub Fil
We caught up with Jakub to talk about spiking neural networks, overcoming the space/AI language gap, and a science fiction book from the man who invented the word “robot.”
FDL: Hi Jacub, where are you from and what is your background?
JF: I’m from Poland, but I’m currently based at the University of Kent. Right now, I’m working on spiking neural networks. In this area of AI, the information is encoded in a form of discrete spikes, or spatio-temporal patterns of activation. This approach to artificial intelligence is modeled more closely to how neurons actually work in the brain. With spiking neural networks, you can more easily encode time-based information and also discrete data. This type of data is used for more discrete applications, such as DNA encoding, which is naturally discrete rather than continuous data.
FDL: As an AI specialist, what is it like working on your first space challenge?
JF: We definitely had some debates at first in terms of terminology. Some of us come from more of a space background and some of us more from AI/machine learning. We didn’t have disagreements, but we didn’t understand each other too well. The first week helped us to build the team and understand each other´s terminology. I’ve been really happy to have the chance to work my teammates, they are all incredible experts in their respective fields.
One big bonus for us is that Marc and Ben on our team have already worked with both earth observation and AI in their own areas of research. Their background has been very helpful for me as they really helped shorten the learning curve in working with satellite imagery.
FDL: Tell me something cool about your FDL experience so far.
Most of AI researchers, myself included, very often work on toy examples. By that I mean we work with pre-processed data sets which are nice and easy to deal with. You just plug it into the model and output something. Here we are working with real data sets from multiple satellites. It’s been great to work with the team to tackle some of the problems that AI encounters when facing the real world.
FDL: Are there any books you’ve particularly enjoyed recently?
I recently read a Czech sci-fi book written in the 1920’s called The Absolute At Large. The author Karel Capek is a really interesting author who actually invented the word robot in one of his previous books. The book deals with the discovery of a nuclear fusion that provide an unlimited source of energy. As a by product, however, these reactors create something called the “Absolute” which affects humans into becoming very religious and nationalistic. It’s particularly interesting since all of this is set in the 1920’s. The world goes to war. China invades Russia and Europe. Japan takes over America who was greatly weakened after a big civil war caused by prohibition. Anyway, let’s just say there are a lot of problems in the end.
About Jakub Fil
Jakub is a recent graduate of the University of Kent (Canterbury, UK), with a first class Bachelor of Science degree in Computer Science and Artificial Intelligence. He currently pursues his PhD there, and is a member of Computational Intelligence research group. His main interests lie in the area of machine learning and neural computation. He is particularly interested in biologically plausible spike-based computation in neural networks employing temporal coding model. Jakub is a keen software engineer who competed in a number of programming competitions, including winning challenges at HackLondon or Imperial College Hackathon. He is an avid space enthusiast, a devoted hitchhiker, and a fan of cycling.