Our Approach to ML

Undoubtedly, it is becoming obvious that artificial intelligence (AI) is increasingly influencing our society and the way we live in it. The completely autonomous car, self-learning smartphones and robots and intelligent programs that help doctors with their diagnoses are no longer just a vision of the future, but are increasingly becoming a technological reality. The machine learning (ML) methods currently used already have a lot in common with the way our human brain works.

Although computers, in contrast to us humans, are able to carry out many highly precise calculations in spaces of any size in a second, human learning is far superior to that of computers in many aspects.

For example, human learning can continuously accumulate knowledge and adapt to randomly changing influences from the environment. Human abilities are not trained for a single specific purpose, but are initially very universal, but can also be specialized through long learning. Nevertheless, in this learning phase, new information can be processed in seconds. In addition, people have the ability to actively learn. We can ask specific questions to get particularly relevant information and we can distinguish between what is important and what is unimportant. Intuitively, we can fill in missing information and have a feeling for the probability of a correct answer to a question. People are not so concerned about making the most accurate prediction possible based on the experiences they have learned; rather, an approximate estimate is often sufficient.

All these capabilities make human learning more effective, efficient and faster compared to machine learning. If machine learning is expanded to include these capabilities, the probability of predicting the correct answers can be increased many times over and the storage and computing requirements can be drastically reduced, so that new areas of application for artificial intelligence can be opened up cost-effectively.

With our advanced ML methods, the valuable treasure trove of data in companies can be leveraged and converted into new information. This makes predictive service and maintenance concepts (predictive maintenance), quality improvements (predictive quality), productivity increases (predictive efficiency) or ML-based anomaly detection, etc. possible.