Robust pruned neural networks via adversarial training
US11562244B2 · kind B2 · utility
1Cited by
3References
23Claims
0Family size
Assignee
Inventors
Key dates
| Filing date | Feb 7, 2019 |
| Grant date | Jan 24, 2023 |
| Priority date | — |
| Expiry date | May 3, 2041 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06N3/094
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
Systems, methods, and computer readable media are described to train a compressed neural network with high robustness. The neural network is first adversarially pre-trained with both original data as well as data perturbed by adversarial attacks for some epochs, then “unimportant” weights or filters are pruned through criteria based on their magnitudes or other method (e.g., Taylor approximation of the loss function), and the pruned neural network is retrained with both clean and perturbed data for more epochs.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.