A proxy model with adaptive constrained optimization enables non-adversarial minimization of the Jeffreys divergence, producing more stable and accurate distribution fitting than MLE or GANs especially in low-data regimes.
Your classifier is secretly an energy based model and you should treat it like one
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.LG 2verdicts
UNVERDICTED 2representative citing papers
MEFA enables exact full-gradient white-box attacks on iterative stochastic purification defenses like diffusion and Langevin EBMs by trading recomputation for lower memory, revealing vulnerabilities missed by approximate-gradient methods.
citing papers explorer
-
Adaptive Symmetrization of the KL Divergence
A proxy model with adaptive constrained optimization enables non-adversarial minimization of the Jeffreys divergence, producing more stable and accurate distribution fitting than MLE or GANs especially in low-data regimes.
-
Memory Efficient Full-gradient Attacks (MEFA) Framework for Adversarial Defense Evaluations
MEFA enables exact full-gradient white-box attacks on iterative stochastic purification defenses like diffusion and Langevin EBMs by trading recomputation for lower memory, revealing vulnerabilities missed by approximate-gradient methods.