Jules Salzinger: One of Reactive Reality’s Deep Learning Heroes
The hero of today’s interview is Jules Salzinger, Reactive Reality’s very enthusiastic Deep Learning expert. He is also the founder of “Transfer Learning”, an online event for Artificial Intelligence (AI) and Deep Learning experts and engineers.
We had the pleasure to interview him and talk about his background, expertise, and current projects in Deep Learning.
Could you please briefly introduce yourself to our readers?
My name is Jules Salzinger, I am 23 and come from a place in France that is close to the border of Luxemburg, Germany, and Belgium. Compared to most other Deep Learning engineers, my background involved a little less research and coding. I studied Engineering at a generalist engineering school in France which gave me the opportunity to learn about as many different scientific subjects as possible, such as chemistry, quantum physics, or mathematics. So that is kind of my background and currently, I am working at Reactive Reality as a Deep Learning Engineer, which I really enjoy.
When and how did you discover Deep Learning?
I discovered it during an internship in a small start-up based in Martinique that produced lamps out of plants and fruit shells. My supervisor told me to check out Artificial Neural Networks and I told him that I did not have enough time and would maybe take a look at it in one or two years. However, I got curious and one week later I spent the whole night researching Artificial Neural Networks. It was the thrill of discovery that got me into Deep Learning.
I started to teach myself as much as possible and tried to focus on AI and Deep Learning in all my projects at university. As I really wanted to dive more into this topic, I decided to do a double degree and went to Taiwan for two years to do so. That is actually what I was doing before Reactive Reality recruited me.
What does the Deep Learning Process look like?
In the beginning we always have a problem that we want to solve. For example, imagine, we have a dataset that contains thousands of pictures of cats and dogs and we need to classify which of those animals is on which picture. We call this a classification problem. Our goal here is to design a neural network that solves this problem for us. To do this, we need a lot of data because we do not code how the neural network will perform the task. Instead, we code how the neural network will learn to perform the task.
Let us compare the neural network to a child. Over time a child grows up, his/her parents show him/her how to perform certain tasks correctly. When he/she does something incorrectly, his/her parents will tell him/her. That is how a child learns. In simple words, it is the same with neural networks. Deep Learning is basically knowing how to design a neural network, analyzing in which cases it is correct or fails, then modifying it so that it learns to perform the task in a better way.
Are you doing what you love professionally?
Absolutely! What I really like to do and what is also a big hobby of mine is teaching others and promoting knowledge about AI and Deep Learning. Approximately three years ago, I started to hold some courses for student associations. I hold these presentations once every two weeks. Also, once a month I give seminars at Reactive Reality. Additionally, I have recently started another project called “Transfer Learning”.
What is Transfer Learning?
Since AI research is currently developing very quickly and it is quite difficult to keep up with everything that is happening in the field, I wanted to create an event that allows experts and engineers to connect, share knowledge and keep in touch with other areas of AI. So, four months ago I took the step and created Transfer Learning, a free online event that takes place every second Tuesday of the month. To get access, interested persons contact us via our website, LinkedIn or email and we send a static link that works for every event.
At Transfer Learning two chosen experts hold a presentation and discuss their findings with the participants afterwards. Furthermore, other topics that have a strong relation to AI and Deep Learning, such as ethics, biology, or philosophy are being discussed. For instance, at one of our last events, we invited a speaker to talk about a very nice method called “AlphaFold” that was recently made in the field of biology due to using AI. Another speaker talked about creating an ethical recommendation algorithm for YouTube which was also very interesting.
Thank you for sharing everything with us, and thank you for the interesting interview, Jules!
Check out our article about AR bringing the in-store experience online.