A kernel-based variant of direct preference optimization and a 100K-pair bias benchmark claim to improve text-to-image safety by separating safe and unsafe image representations during training.
Adding integrable defects to the Boussinesq equation
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
The purpose of this paper is to extend the store of models able to support integrable defects by investigating the two-dimensional Boussinesq nonlinear wave equation. As has been previously noted in many examples, insisting that a defect contributes to energy and momentum to ensure their conservation, despite the presence of discontinuities and the explicit breaking of translation invariance, leads to sewing conditions relating the two fields and their derivatives on either side of the defect. The manner in which several types of soliton solutions to the Boussinesq equation are affected by the defect is explored and reveals new effects that have not been observed in other integrable systems, such as the possibility of a soliton reflecting from a defect or of a defect decaying into one or two solitons.
citation-role summary
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
REJECT 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
DETONATE: A Benchmark for Text-to-Image Alignment and Kernelized Direct Preference Optimization
A kernel-based variant of direct preference optimization and a 100K-pair bias benchmark claim to improve text-to-image safety by separating safe and unsafe image representations during training.