Martin Josifoski

I am a PhD student at EPFLūüá®ūüá≠adviced by Robert West. Previously, I received a M.Sc. in Data Science from EPFL where I worked as a research assistant at The Data Science Lab; I completed my M.Sc. thesis at the Learning & Adaptive Systems Group, ETHZ.

I spent some time at Microsoft Research, Redmond and FAIR, London.

My research is supported by a fellowship from the Swiss Data Science Center and grants/awards from Microsoft and Google.

My long-term research focuses on the development of a high-capable collaborative personal AI assistant that adapts to our evolving idiosyncratic needs, goals, and drives and can be trusted to act in line with our values and interests. Currently, I am working on enhancing collaboration and, since recently, memory in support of learning, adaptation, and inference.

I am the lead contributor to aiflows, a library for structured interactions of AI systems, tools, and humans.

news

Dec, 2023 Our work on PAC-Bayesian Meta-Learning was accepted for publication in JMLR!
Nov, 2023 Gave a talk at Swiss Data Science Center: Reasoning and Collaborating AI!
Oct, 2023 Gave a talk at Google Research, Z√ľrich: Flows: Building Blocks of Reasoning and Collaborating AI!
Oct, 2023 Two papers accepted at EMNLP‚Äô23 ūüöÄ [link] [link]
Aug, 2023 Paper out: ūüĆä Flows: Building Blocks of Reasoning and Collaborating AI
Jun, 2023 Excited to start my internship at Microsoft Research, where I will be working with Eric Horvitz, Adam Fourney, and Gagan Bansal!
Jun, 2023 BLINK passed 1K stars ‚≠ź on GitHub

selected publications

  1. Flows: Building Blocks of Reasoning and Collaborating AI
    Martin Josifoski, Lars Klein, Maxime Peyrard, Yifei Li, Saibo Geng, and 5 more authors
    Under Review, 2023
  2. Exploiting Asymmetry for Synthetic Training Data Generation: SynthIE and the Case of Information Extraction
    Martin Josifoski, Marija Sakota, Maxime Peyrard, and Robert West
    EMNLP, 2023
  3. PAC-Bayesian Meta-Learning: From Theory to Practice
    Jonas Rothfuss, Martin Josifoski, Vincent Fortuin, and Andreas Krause
    JMLR, 2023
  4. GenIE: Generative Information Extraction
    Martin Josifoski, Nicola De Cao, Maxime Peyrard, Fabio Petroni, and Robert West
    NAACL, 2022
  5. Scalable Zero-shot Entity Linking with Dense Entity Retrieval
    Ledell Wu, Fabio Petroni, Martin Josifoski, Sebastian Riedel, and Luke Zettlemoyer
    EMNLP, 2020