Machine learning is among the most successful research areas in the recent times, particularly since the widespread of big data. In this context, deep learning is a term that encapsulates a family of related supervised and unsupervised learning methods, based on artificial neural networks. Deep learning has recently been associated with revolutionary Artificial Intelligence achievements, ranging from “close-to-human” speech and image recognition performances, up to “super-human” game playing results. Throughout this course, students will have the opportunity to understand the building blocks of neural networks. The curriculum starts by introducing supervised learning concepts and incrementally dives into the peculiarities of learning the parameters of neural networks through back-propagation. Specific architectures, such as the Convolutional Neural Networks will be covered, as well as different lectures of network regularization strategies. Furthermore implementation techniques involving GPU-based optimization will be explained. The students are expected to master the necessary knowledge that will empower them to apply Deep Learning in real-life problems.
Lecturer: Prof. Dr. Dr. Lars Schmidt-ThiemeTrainer: Randolf Scholz