Welcome page

Prof. Dr. Grabocka has moved to the University of Technology Nurnberg, where he is leading the Machine Learning Lab. As a result, the RELEA lab at the University of Freiburg is not active anymore. Please follow our recent work at: https://scholar.google.com/citations?user=KRy27XcAAAAJ&hl=de&authuser=1

The Representation Learning Lab (RELEA) led by J.-Prof. Dr. Josif Grabocka was founded in 01.12.2019 and focuses on exploring Deep Learning representations to tackle diverse Machine Learning tasks arising in practical data-driven application domains. In particular, the lab is focused on Neural Architecture Search, Hyper-Parameter Optimization, as well as designing end-to-end architectures for sequential data (time-series) and Recommender System prediction tasks.

Recent News

21.09.2023 Our paper "Power Laws for Hyperparameter Optimization" authored by Arlind Kadra, Maciej Janowski, Martin Wistuba and Josif Grabocka was accepted for publication at the Thirty-Seventh Conference on Neural Information Processing Systems (NeurIPS 2023).

22.05.2023 Our paper "Deep Pipeline Embeddings for AutoML" authored by Sebastian Pineda Arango and Josif Grabocka was accepted at the ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2023).

22.01.2023  Three RELEA research papers were accepted for publication at the International Conference on Learning Representations (ICLR 2023).
14.09.2022  Our latest work, Supervising the Multi-Fidelity Race of Hyperparameter Configurations is accepted in the Thirty-sixth Conference on Neural Information Processing Systems (NeurIPS 2022).
08.03.2022  RELEA will participate in the project “Responsible and Scalable Learning for Robots Assisting Humans” (ReScaLe) starting on 1 May 2022. Prof. Dr. Grabocka will lead a work package on meta-learning with meta-features for Reinforcement Learning.

11.10.2021 We release HPO-B, a new benchmark for Black-Box HPO that is additionally published as a research paper at the Conference on Neural Information Processing Systems, Datasets and Benchmarks Track (NeurIPS 2021). For more, read here.

28.09.2021 Our newest paper Well-tuned Simple Nets Excel on Tabular Datasets is accepted in the Thirty-fifth Conference on Neural Information Processing Systems (NeurIPS 2021).

13.08.2021 Prof. Grabocka gave an invited talk titled “Deep Learning for Tabular Datasets” at the Freiburg Center for Data Analysis and Modeling’s seminar on "Data Analysis and Modeling".

12.01.2021 Our newest paper Few-Shot Bayesian Optimization with Deep Kernel Surrogates is accepted in the Ninth International Conference on Learning Representations (ICLR 2021).