The group at the Frankfurt Institute for Advanced Studies (FIAS) is contributing to research area C. We are developing machine- and deep-learning tools for the CBM experiment, including data analysis and detector design. Deep Convolutional Neural networks (DCNN) are applied to detect the QCD phase transition from realistic simulations of the CBM experiment. Acceptance and efficiency limitations of the detector will be considered in a hybrid simulation for heavy ion collisions in CBM. In a final step the developed ML/DL tools can then be used on real data within the CBMroot. Furthermore, both supervised and unsupervised machine-learning techniques will be developed for fast physics extraction and event class selection, which is important for the online selection of ‘interesting physics’ events from the large event rate at the CBM experiment.