Machine learning methods can solve a variety of tasks (such as image classification) that are too complicated to be solved satisfyingly with traditional approaches. By analyzing training data sets representing the problem at hand machine learning methods can draw generalizations resulting in robust solutions to even challenging tasks.

Problems in which labels must be found for certain data (such as finding the object present in an image or predicting the house price from certain properties of the house) are often too complicated to be solved by deterministic methods. Supervised learning can be used to find a satisfying approximate solution to such problems by manually labeling some of the data and generalizing from that hand - labeled data.

Unsupervised learning can help to uncover hidden structure (i.e. latent variables) in unlabeled data which can be useful to perform tasks such as clusterization, dimensionality reduction or anomaly detection.

We aid companies in finding out whether it is helpful to utilize Machine learning methods to their respective problems or whether they can be solved by other means. If it turns out that Machine learning is indeed well-suited for the task, we can develop corresponding solutions.