Artificial Intelligence

Ein Anwendungsfall für Künstliche Intelligenz: Automatische Erkennung von Szenen.
© Fraunhofer IPA; Photo: Rainer Bez

Artificial intelligence has already found its way into production in a variety of forms: From intelligent learning processes for robots to automatic error detection. In contrast, the use of quantum computers is only just beginning, but is also considered a key technology. While quantum computing and machine learning have long been considered separate areas of research, it is now evident that the two can complement each other highly effectively.

When machines become smart: Artificial Intelligence and Machine Learning

Everyone hears about Artificial Intelligence (AI) and Machine Learning (ML). But what do the two terms essentially mean? We talk about artificial intelligence when machines are enabled to solve problems that would require human intelligence. Closely related to this is machine learning, i.e., machines or robots generating knowledge from experience and consequently solving tasks on the basis of learned patterns and regularities.

According to a Bitkom study from 2019, 49% of industrial companies consider it very likely or somewhat likely that artificial intelligence will disruptively change existing business models in the context of Industry 4.0. The companies surveyed perceive the most important benefits of artificial intelligence to be increased productivity (47%), predictive maintenance (39%), optimization of production and manufacturing processes (33%), and increased product quality (25%).

Machine Learning as a subfield of Artificial Intelligence has made enormous progress over the last decade. The most common methods to enable machines to deal with complex learning tasks are logistic regression (63.5%), decision trees (49.9%), random forests (46.3%) and neural networks (37.6%), according to a Fraunhofer study published in 2018. Today, applications trained with ML are not only capable of analyzing image, video and text data, but also, for example, creating text translations, answering emails, composing music or producing images.

A »quantum leap« for Machine Learning

Quantum computers, i.e. computers that operate according to the laws of quantum mechanics, have only recently begun to receive attention in connection with machine learning. The commonality is that quantum computers detect subtle patterns in large amounts of data, just as is done in the process of Machine Learning. At the moment, there is still a lot of research and development to be done in the field of quantum computing. However, it is already foreseeable that this technology could also open up new possibilities in the context of ML.

At the moment, Machine Learning is reaching its limits in applications where it is necessary to evaluate continuously accumulating large amounts of data, from which the machine is then supposed to learn in real time. Quantum computing could solve this problem in the medium term, as numerous parallel calculations are possible with a quantum computer.

Our partners

Fraunhofer IPA Logo
Logo Fraunhofer IAO
Fraunhofer IBP Logo