A concept representation of a machine learning model
By deriving a concept representation of ML models using feature vector summaries, the limitations of current explanation methods are overcome, enabling efficient and interpretable evaluation and improvement of ML models.
WO2026125313A1PCT designated stage Publication Date: 2026-06-18FRAUNHOFER GESELLSCHAFT ZUR FORDERUNG DER ANGEWANDTEN FORSCHUNG EV
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- FRAUNHOFER GESELLSCHAFT ZUR FORDERUNG DER ANGEWANDTEN FORSCHUNG EV
- Filing Date
- 2025-12-09
- Publication Date
- 2026-06-18
Smart Images
Figure EP2025085997_18062026_PF_FP_ABST
Abstract
A concept representation of a machine learning model, the concept representation comprising, for each of a plurality of model portions of the machine learning model, a respective feature vector representation, is derived by deriving, for a model portion of the plurality of model portions, a set of data samples, which data samples are representative of a concept associated with the model portion, and subjecting the data samples to a further machine learning model, e.g., a foundation model, to obtain the feature vector representation of the model portion.
Need to check novelty before this filing date? Find Prior Art