Jan Melchior

Dr.-Ing. Jan Melchior is a Researcher at the Institute for Neural Computation within the Faculty of Computer Science at Ruhr-University Bochum, Germany. He has held the position of Doctoral Researcher since 2012 in Prof. Dr. Laurenz Wiskott's group. Melchior completed his education at Ruhr-University Bochum, earning a Master of Computer Science (2012, thesis on Gaussian-Binary Restricted Boltzmann Machines) and a Bachelor of Computer Science (2009).

His research focuses on unsupervised machine learning, emphasizing probabilistic models like Boltzmann machines and deep learning. Key interests include developmental-online learning, growing systems, and semi-supervised techniques, which he considers vital for generalization and knowledge integration. His work bridges machine learning with neuroscience, particularly in sequence memory, image statistics, and neural modeling.

Melchior has supervised multiple students, including bachelor/master theses and study projects. He taught as an assistant for Machine Learning: Unsupervised Methods (2017) and led lab courses for Scientific Computing with Python (2015–2017). He developed PyDeep, a Python library for unsupervised learning (e.g., PCA, RBM, autoencoders).