The output error from fully convex-relaxed neural network verification grows exponentially with depth and linearly with input radius, with misclassification probability showing step-like dependence on radius.
Huang and Duligur Ibeling and Kyle Julian and Christopher Lazarus and Rachel Lim and Parth Shah and Shantanu Thakoor and Haoze Wu and Aleksandar Zeljic and David L
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
The Cost of Relaxation: Evaluating the Error in Convex Neural Network Verification
The output error from fully convex-relaxed neural network verification grows exponentially with depth and linearly with input radius, with misclassification probability showing step-like dependence on radius.