Synthetic Data from Diffusion Models Improves ImageNet Classification
This work addresses the challenge of enhancing discriminative tasks like image classification for AI researchers and practitioners, but it is incremental as it applies an existing generative method to a new application.
The authors tackled the problem of improving ImageNet classification by using synthetic data from fine-tuned diffusion models for generative data augmentation, resulting in significant accuracy improvements over ResNet and Vision Transformer baselines, with concrete metrics such as SOTA FID of 1.76 and Classification Accuracy Scores up to 69.24.
Deep generative models are becoming increasingly powerful, now generating diverse high fidelity photo-realistic samples given text prompts. Have they reached the point where models of natural images can be used for generative data augmentation, helping to improve challenging discriminative tasks? We show that large-scale text-to image diffusion models can be fine-tuned to produce class conditional models with SOTA FID (1.76 at 256x256 resolution) and Inception Score (239 at 256x256). The model also yields a new SOTA in Classification Accuracy Scores (64.96 for 256x256 generative samples, improving to 69.24 for 1024x1024 samples). Augmenting the ImageNet training set with samples from the resulting models yields significant improvements in ImageNet classification accuracy over strong ResNet and Vision Transformer baselines.