Embedded Files
Representations in generative AI: Causal methods, images, music, and language.
Recent progress in generative AI has not only achieved generation of highly realistic synthetic data, but also strives to allow for manipulation of high-level properties of the data, with significant potential for applications. Building useful representations is a core ingredient to be able to perform such meaningful manipulations: this goes beyond the mere representation of statistical information, and also includes aspects of causality. The seminar will explore the theoretical and empirical properties of these representations.
Recent progress in generative AI has not only achieved generation of highly realistic synthetic data, but also strives to allow for manipulation of high-level properties of the data, with significant potential for applications. Building useful representations is a core ingredient to be able to perform such meaningful manipulations: this goes beyond the mere representation of statistical information, and also includes aspects of causality. The seminar will explore the theoretical and empirical properties of these representations.
Organization: Students will be able to choose from by ranking papers on a list available via the moodle webpage from Thursday, December 14th 2023 until Monday, January 1st 2024. Papers will be assigned based on this ranking. The seminar will take place on February 8-9th in CAB H52 and starts at 9:30am. We expect participants to attend the full duration of the seminar and participate actively by asking questions and contributing to the discussion.
Zoom: https://ethz.zoom.us/j/65879537239
Schedule:
List of relevant papers:
Organizers
Organizers
MPI for Intelligent Systems
MPI for Intelligent Systems
MPI for Intelligent Systems
MPI for Intelligent Systems/ETH Zürich
Page updated
Google Sites
Report abuse