GABE: Groupwise Affine Basis Encoding
Introduce GABE (Groupwise Affine Basis Encoding) - a decomposition method that represents neural network weights as an addressable memory system. For any group of similar layers, we extract three interpretable components: a shared mean weight (long-term memory), a low-rank basis of variations (address space), and per-layer coefficients (pointers).
Authors
- Dmitry F. Feklin
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-02-05
- DOI
- https://doi.org/10.5281/zenodo.18488810
- Primary Topic
- Adversarial Robustness in Machine Learning
- Type
- article
- Field-Weighted Citation Impact
- 0.00