Closest decision-boundary (or 'flip') points are computed for a CIFAR-10 network, showing Taylor approximations underestimate distances and that robustness measures should use these exact points.
Matrix Computations
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Investigating Decision Boundaries of Trained Neural Networks
Closest decision-boundary (or 'flip') points are computed for a CIFAR-10 network, showing Taylor approximations underestimate distances and that robustness measures should use these exact points.