Abstract
A computing system can obtain input data containing semantic information. The computing system can access conditioning data configured to cause a generative model to generate modifications to the input data which alter the input data to satisfy a data refinement constraint while preserving the semantic information. The computing system can process the input data with the generative model conditioned on the conditioning data to generate, as an output of the generative model, refined data. According to an aspect of the present disclosure, the refined data can both satisfy the data refinement constraint and also contain the semantic information.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Jiang, Minqi; Madeira Araújo, João; Ellsworth, William; Gooding, Sian Heather; and Grefenstette, Edward, "Generative Data Refinement with Semantic Preservation", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10032