PRISM: Diversifying Dataset Distillation by Decoupling Architectural Priors
We introduce PRISM, a framework that disentangles architectural priors for dataset distillation, outperforming single-teacher setups.
Arundhati Shanbhag.
DFKI
RPTU Kaiserslautern–Landau
Co-authored Publications: 4
SubZeroCore: A Submodular Approach with Zero Training for Coreset Selection
We introduce SubZeroCore, a novel, training-free coreset selection method that integrates submodular coverage and density into a single, unified objective.
HyperCore: Coreset Selection under Noise via Hypersphere Models
We present HyperCore, a lightweight adaptive coreset selection framework designed for noisy environments. HyperCore utilizes per class hypersphere models and adaptively selects pruning thresholds.
Just Leaf It: Accelerating Diffusion Classifiers with Hierarchical Class Pruning
We speed up diffusion classifiers by utilizing a label hierarchy and pruning unrelated paths.