معرفی
Erik Schultheis is a Doctoral Student and Visitor (Faculty) in the Department of Computer Science at the School of Science. His research focuses on extreme multi-label classification, algorithm design, and optimization challenges in machine learning. He holds a Master of Science in Physics from Georg-August-Universität Göttingen (2019).
Key research interests include scalable classification systems, label correlation learning, and addressing long-tail and missing label problems in large-scale datasets. His work emphasizes practical solutions for commodity hardware limitations and memory efficiency in training models with millions of labels.
Notable contributions include the Dismec++ software tool for extreme classification, and he has presented at conferences such as NeurIPS, ICML, and KDD. His 2021 NeurIPS Outstanding Reviewer Award highlights his peer-review contributions to the field.
Recent publications explore dynamic sparsity in large output spaces, online algorithm generalizations, and label calibration in extreme classification scenarios. His research bridges theoretical algorithm development with practical implementation challenges in real-world systems.
