FVL-FP combines cross-layer fair prompts, orthogonal projection off demographic subspaces, and fairness-weighted prompt fusion to reduce group bias in federated vision-language models.
An unexplored valley of binary packing: The loose jamming state
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
We present a theoretical prediction on random close packing factor \phi_RCP^b of binary granular packings based on the hard-sphere fluid theory. An unexplored regime is unravelled, where the packing fraction \phi_RCP^b is smaller than that of the mono-sized one \phi_RCP^m, i.e., the so-called loose jamming state. This is against our common perception that binary packings should always reach a denser packing than mono-sized packings at the jamming state. Numerical evidence further supports this prediction and confirms the regime location in the size ratio and mole fraction (R_r-X_s) space, where the size ratio R_r is close to 1, and the mole fraction of the smaller sphere X_s close to 0. This extreme regime remains unreported in existing literature, yet significant for our fundamental understanding of binary packing systems.
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Mitigating Group-Level Fairness Disparities in Federated Visual Language Models
FVL-FP combines cross-layer fair prompts, orthogonal projection off demographic subspaces, and fairness-weighted prompt fusion to reduce group bias in federated vision-language models.