NCFM: Dataset Distillation with Neural Characteristic Function
Overview NCFM is the official PyTorch implementation of the paper "Dataset Distillation with Neural Characteristic Function: A Minmax Perspective" presented at CVPR 2025. This method focuses on dataset distillation, allowing for efficient model training with fewer data samples.
Key Features:
- PyTorch Implementation: Provides a robust framework for researchers and developers.
- Minmax Perspective: Unique approach to dataset distillation that optimizes data usage.
- Comprehensive Documentation: Step-by-step guide for installation, pretraining, and evaluation.
Benefits:
- Enables quicker experimentation and model development with reduced datasets.
- Facilitates state-of-the-art results in machine learning tasks using fewer resources.
Getting Started:
- Clone the repository:
git clone https://github.com/gszfwsb/NCFM.git - Install dependencies:
pip install -r requirements.txt - Follow the provided examples for CIFAR-10 and CIFAR-100.