When did machine learning become popular?

What does a machine learning developer do?

Machines are becoming increasingly intelligent because there are people who teach them to learn. We clarify with Damian Borth from the DFKI what these software experts have to bring with them and why they have to be inquisitive themselves.

Photo: panthermedia.net/ Kittipong Jirasukhanont

If Siri understands our speech pattern in our smartphone, the auto-completion of chat programs recognizes what we intend to type earlier and earlier, or our e-mail program reliably separates spam from desired messages, then we are dealing with machine learning. We are already encountering the sub-discipline of artificial intelligence at every turn and making rapid progress in the process. But what does the job description of a machine learning developer look like? We spoke to Damian Borth from the German Research Center for Artificial Intelligence (DFKI) in Kaiserslautern about this.

ingenieur.de: Mr. Borth, how do you assess the state of research in artificial intelligence and machine learning?

Damian Borth is a computer scientist and director of the Deep Learning Competence Center at the German Research Institute for Artificial Intelligence (DFKI) in Kaiserslautern. He wrote his doctorate at the Department of Computer Science at the TU Kaiserslautern and at the Competence Center Multimedia Analysis and Data Mining (MADM) at the DFKI.

Source: DFKI

Damian Borth: As very good. Especially in the area of ​​deep learning, i.e. the area of ​​machine learning based on deep, neural networks, there has been great progress since 2012. The technology is very successful when it comes to vision, translation and speech recognition. Today, for example, it is predictable which emotions a picture will trigger and what popularity it will achieve.

Do research and industry have enough developers for machine learning?

Not at the moment, that's why they're so popular. The data analyst is one of the most important professions of the future. It is not for nothing that it is already considered the "sexiest job of the century."

Already today we are surrounded by algorithm-based models of artificial intelligence. For example, when we interview a search engine, use facial recognition for pictures or inquire about a loan in a bank and our rating is called up. All of this is enriched with artificial intelligence, even if we are not always aware of it. And the areas of application are becoming more and more. Artificial intelligence is considered to be the computer science of the future.

That sounds like a crisis-proof workplace. What does someone have to be able to do to develop the necessary algorithms and simulations?

A degree in computer science with a focus on mathematics or, conversely, mathematics with a focus on computer science is ideal. This is so important because the development of algorithms is very formal. A neuroscientist who understands how the human brain is structured and can contribute this knowledge is also conceivable.

But it is possible for me to get on board, for example from physics or other disciplines.

Then where do I get in with this knowledge?

Wherever you want. Data science is the best discipline to make a career in business. The financial industry needs developers, retail, mechanical engineering with its robot technology and human-machine communication. Everything is moving towards the digital sphere. We are still at the beginning of transforming data and information into service products. The rethinking that information can be used is only just beginning.

And then do I spend my professional life programming algorithms?

But no. Collaboration is particularly important in research. The exchange within working groups or with students. The more people involved in a project, the more feedback and advice is needed. In addition, research is advancing at such a pace that many scientific papers are being published. There is a lot to read, learn, and discuss. But of course there are also phases in front of the computer when you try to implement your networks and get them up and running.

That sounds like a spirit of optimism and exciting. What is the situation outside of research in a company?

That depends on how much you're allowed to experiment. Smart companies rely on two-speed IT. On the one hand the IT, which accompanies production and processes in practice, and on the other hand isolated islands where new things can be tried out. Once a developer has found the company that suits them, even more collaboration is required than in research: across all disciplines, from product development to the legal department and marketing. For this we need people who can communicate across all areas, levels and levels of detail.

What other properties are important?

A good team can handle every archetype from the genius to the loner - as long as it is only represented once per team. But you will be successful if you are able to find a narrative around the technology, an explanation of where it will lead and what good it will bring for companies. Technology is always a promise to the future. Those who recognize the advantages, can set up a lab, look for and find use cases and also have an entrepreneurial spirit, will go their own way.

Note from the editors: Together with acatech - the German Academy of Engineering Sciences, the DFKI has launched the first German-language, free online course for machine learning.


More job profiles:

Engineers and IT - jobs with a future

What does a test manager do?

What does an embedded systems engineer do?

How do you become a BIM manager?


A contribution by: