VectorDefense: Vectorization as a Defense to Adversarial Examples
By Vishaal Munusamy Kabilan, Brandon L. Morris, Anh Totti Nguyen. 2018
22 citations recorded by the source corpus.
Training deep neural networks on images represented as grids of pixels has brought to light an interesting phenomenon known as adversarial examples. Inspired by how humans reconstruct abstract concepts, we attempt to codify the input bitmap image into a set of compact, interpretable elements to avoid being fooled by the adversarial structures. We take the first step in this direction by experimenting with image vectorization as an input transformation step to map the adversarial examples back into the natural manifold of MNIST handwritten digits. We compare our method vs. state-of-the-art input transformations and further discuss the trade-offs between a hand-designed and a learned transformation defense.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Wieland Brendel
- Matthias Bethge
- Jonas Rauber
- Lukas Schott
- Florian Tramèr
- Jörn-Henrik Jacobsen
- Nicholas Carlini
- Richard S. Zemel
In Computer Vision and Pattern Recognition
- Wieland Brendel
- Matthias Bethge
- Jonas Rauber
- Lukas Schott
- Luc Van Gool
- Trevor Darrell
- Andrea Vedaldi
- Ming-Hsuan Yang
Citations in the corpus
- Towards the first adversarially robust neural network model on MNIST (2018, background): “• Preprocessing: Inputs or hidden activations are quantized, projected into a different representation or are otherwise preprocessed (Buckman et al., 2018; Guo et al., 2018; Kabilan et al., 2018).”