Zero-shot Generation of Training Data with Denoising Diffusion Probabilistic Model for Handwritten Chinese Character Recognition.
Dongnan GuiKai ChenHaisong DingQiang HuoPublished in: ICDAR (2) (2023)
Keyphrases
- denoising
- probabilistic model
- handwritten chinese character recognition
- training data
- nonlinear diffusion
- diffusion processes
- image denoising
- bayesian networks
- noisy images
- total variation
- learning algorithm
- anisotropic diffusion
- language model
- test data
- training set
- data sets
- denoising algorithm
- domain knowledge
- supervised learning
- wavelet domain
- noise removal
- image processing
- wavelet denoising
- learned from training data
- diffusion process
- classification accuracy
- training process
- object categories
- test set
- training samples
- natural images
- generalization error
- classification models
- diffusion equation
- training instances
- diffusion model
- tensor field
- unlabeled data
- generative model
- neural network