Johannes Ackermann

prof_pic.jpg

I’m a third-year PhD student at the University of Tokyo, working on Reinforcement Learning supervised by Professor Masashi Sugiyama.

I’m particularly interested in how we can deal with changing or complicated transition and reward functions in RL, as encountered in Multi-Agent interaction, Multi-Task settings or due to Non-Stationarity.

I’m also employed by RIKEN AIP as a part-time researcher and supported by a Japanese government MEXT scholarship, and a Microsoft Research Asia D-CORE grant.

I interned at Preferred Networks and previously worked as a DSP researcher in Huawei’s Munich Research Center, working on applied Machine Learning. Before that, I received my B.Sc. and M.Sc. in Electrical Engineering and Information Technology from the Technical University of Munich in 2018 and 2021. I wrote my Master’s Thesis about Multi-Task RL in ETH Zurich’s Distributed Computing group.

news

May 17, 2024 Our work on Offline Reinforcement Learning from Datasets with Structured Non-Stationarity was accepted at RLC 2024:tada:
Dec 05, 2022 I presented our work High-Resolution Image Editing via Multi-Stage Blended Diffusion from my internship in PFN at the NeurIPS Machine Learning for Creativity and Design Workshop :tada:
Apr 20, 2022 I posted a blog-entry detailing how I built a Text to Image web app!

latest posts

publications

  1. RLC
    nonstationary_dataset.png
    Offline Reinforcement Learning from Datasets with Structured Non-Stationarity
    Johannes Ackermann, Takayuki Osa, and Masashi Sugiyama
    In Reinforcement Learning Conference (RLC) 2024 , Aug 2024
  2. Leveraging Domain-Unlabeled Data in Offline Reinforcement Learning across Two Domains
    Soichiro Nishimori, Xin-Qiang Cai, Johannes Ackermann, and Masashi Sugiyama
    Aug 2024
    Preprint
  3. msbd_overview.jpg
    High-Resolution Image Editing via Multi-Stage Blended Diffusion
    Johannes Ackermann, and Minjun Li
    In NeurIPS Machine Learning for Creativity and Design Workshop 2022 , Dec 2022
  4. Unsupervised Task Clustering for Multi-Task Reinforcement Learning
    Johannes Ackermann, Oliver Richter, and Roger Wattenhofer
    In ECML-PKDD 2021 , Sep 2021
  5. Convolutional Neural Network Based Blind Estimation of Generalized Mutual Information for Optical Communication
    Johannes Ackermann, Maximilian Schädler, and Christian Blümm
    In European Conference on Optical Communication (ECOC) , Dec 2020
  6. estimation_matd3.png
    Reducing Overestimation Bias in Multi-Agent Domains Using Double Centralized Critics
    Johannes Ackermann, Volker Gabler, Takayuki Osa, and Masashi Sugiyama
    In Deep Reinforcement Learning Workshop at NeurIPS , Dec 2019