We present a training system, which can provably defend significantly larger neural networks than previously possible, including ResNet-34 and DenseNet-100. Our approach is based on differentiable abstract interpretation and introduces two novel concepts: (i) abstract layers for fine-tuning the precision and scalability of the abstraction, (ii) a flexible domain specific language (DSL) for describing training objectives that combine abstract and concrete losses with arbitrary specifications. Our training method is implemented in the DiffAI system.
A Provable Defense for Deep Residual Networks
A training system using differentiable abstract interpretation defends larger neural networks by fine-tuning abstraction and combining abstract and concrete losses.
- Year
- 2019
- Venue
- arXiv 2019
- Authors
- 3
- Hosting
- Abstract onlyARXIV-DEFAULT
Cite
Notes
Only stored in your browser.
Attribution
- Abstract & full text
- arxiv.org/abs/1903.12519v2ARXIV-DEFAULT
- TL;DR
- Semantic Scholar