Artificial intelligence and neural networks as innovation turbocharger 4 minutes spent reading
Innovation

Artificial intelligence and neural networks as innovation turbocharger

Viktoria Steininger
Holds editorial responsibility for blog topics, is researching and writing articles. Her stories give insights into the world of the voestalpine Group.

Digitization, neural networks, and deep learning are the prerequisites for artificial intelligence and machine learning. The first projects at voestalpine show that these are all not just future visions.

sepp hochreiterSimply put, artificial intelligence (AI) is the ability of computer systems to work on problems themselves. This means that they can recognize the environment, plan, reason logically—and learn. Similar to the human brain that can anatomically change its nerve cells to trigger the learning process, an artificial neural network can also “learn”—by recognizing repeated patterns. As AI pioneer Sepp Hochreiter (Professor at the Johannes Kepler University of Linz) explained at the first voestalpine Digitization Day, such systems require three things:Simply put, artificial intelligence (AI) is the ability of computer systems to work on problems themselves. This means that they can recognize the environment, plan, reason logically—and learn. Similar to the human brain that can anatomically change its nerve cells to trigger the learning process, an artificial neural network can also “learn”—by recognizing repeated patterns. As AI pioneer Sepp Hochreiter (Professor at the Johannes Kepler University of Linz) explained at the first voestalpine Digitization Day, such systems require three things:

  • Very fast computers
  • Extensive data sets
  • Neural networks
"Digitization prepares the way for artificial intelligence. Digitization is like electricity before there were computers."
Sepp Hochreiter (JKU Linz)

Today, there are many application examples of artificial intelligence and neural networks, e.g. in medical diagnosis, computer algorithms (Google PageRank and search results), and autonomous driving. But how can neural methods be used in other areas, such as in the voestalpine Group?

Wide range of applications for voestalpine

According to AI expert Hochreiter, artificial intelligence can be used almost everywhere in voestalpine, for example to optimize production processes. AI systems that are very good at recognizing faces, for instance, would also be good at recognizing (faulty) patterns on product surfaces when producing steel strip, flat steel, etc. Together with selected external partners and internal departments, the mechatronics research department is already working on specific tasks that use deep learning methods. And Steel Division processes are being examined to see which ones meet the prerequisites for successfully using deep learning applications. Another exciting example can be found in the High Performance Metals Division at BÖHLER Edelstahl in Kapfenberg where they are testing the use of artificial neural networks (ANNs) in inductive hardening of steel bars. Different parameters, such as the voltage of the induction coils and the feed speed of the bars, need to be controlled. They are investigating how ANNs can evaluate and control these parameters to improve the quality of the bars. The goal is a self-optimizing process as well as improving and accelerating knowledge transfer between research and production.

kuenstliche-intelligenzDespite the highly complex technology, when it comes to artificial intelligence, people are more important than ever before since these systems must first be trained and then continually optimized. Sepp Hochreiter has a clear response to the question of whether intelligent machines will replace employees in the future: “No, definitely not. I need the teacher. I need someone who can train the machine. I need someone who has experience, someone who has been with the company for a long time and can tell the systems what has gone wrong. Artificial intelligence is simply a way to support people.”

Viktoria Steininger