Dimensions overall score 7.0

GitHub Code Pulse

No public code linked for this paper yet.

• The paper establishes a theoretical framework proving the equivalence between dataset distillation and distribution matching.technical 90%
• The proposed Dataset Concentration (DsCo) framework achieves state-of-the-art performance for low data volumes.result 85%
• DsCo can reduce a dataset size by nearly half with no performance degradation in high data volume scenarios.result 90%
• The method is applicable in both data-accessible and data-free scenarios.method 90%
• Existing diffusion-based dataset distillation methods lack theoretical justification.limitation 85%
• The paper reveals an inherent efficiency limit in the dataset distillation paradigm.technical 80%
• The Doping augmentation method mixes selected original samples with synthetic samples to overcome the efficiency limit of dataset distillation.method 85%
• In data-free scenarios, DsCo outperforms all existing data-free dataset distillation methods.result 85%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card