Recognize patterns and structures in data
Machine learning (ML), a sub-discipline of artificial intelligence (AI), aims to enable systems to recognize patterns in data. Typical use cases include classification and prediction, which are applied in various fields such as image, video and speech recognition. By using ML, complex data structures can be analyzed and used for precise predictions.
The machine learning process involves several stages: from data preparation (harmonization, transformation, and enrichment) and selection of explanatory variables to training and evaluation of the models. When implemented in challenging real-time environments, the performance of the models is critical. While many ML models, especially neural networks, are often considered "black boxes" and difficult to interpret, the field of explainable artificial intelligence (XAI) is an active research discipline aimed at improving the comprehensibility and interpretability of ML models, which is also of great importance for INFORM's applications. Deep Learning, a special class of neural networks, has established itself as a powerful method in the field of machine learning.
Within INFORM, new applications of machine learning are regularly uncovered, tested, and implemented. The results of the ML models often bring added value to the companies using them. However, they can also be used to provide the search and optimization algorithms from operations research with an even better data basis for current calculations. Examples of customer successes that are being actively used include delivery time forecasts for ordered goods, the recognition of new fraud patterns in the financial sector, or the evaluation of incoming and outgoing containers at the port in order to stack them in the right place for the expected means of transport.