Training Unbiased Diffusion Models From Biased Dataset

Authors: Yeongmin Kim, Byeonghu Na, Minsang Park, JoonHo Jang, Dongjun Kim, Wanmo Kang, Il-chul Moon

ICLR 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experimental evidence supports the usefulness of the proposed method, which outperforms baselines including time-independent importance reweighting on CIFAR-10, CIFAR-100, FFHQ, and Celeb A with various bias settings.
Researcher Affiliation Collaboration Yeongmin Kim1 , Byeonghu Na1, Minsang Park1, Joon Ho Jang1, Dongjun Kim1, Wanmo Kang1, Il-Chul Moon1,2 (...) 1KAIST, 2Summary.AI
Pseudocode Yes Algorithm 1: Discriminator Training algorithm (...) Algorithm 2: Score Training algorithm with TIW-DSM
Open Source Code Yes Our code is available at https://github.com/alsdudrla10/TIW-DSM.
Open Datasets Yes We consider CIFAR-10, CIFAR-100, FFHQ, and Celeb A datasets, which are commonly used for generative learning.
Dataset Splits No The paper does not explicitly provide the specific percentages or counts for training/validation/test dataset splits from the observed dataset (Dbias) to reproduce the experiment's data partitioning.
Hardware Specification Yes Table 7 shows the computational costs measured using RTX 4090 4 cores in the CIFAR-10 experiments.
Software Dependencies No The paper mentions using PyTorch and following procedures from EDM, but does not provide specific version numbers for these or other software dependencies.
Experiment Setup Yes Table 6: Training and sampling configurations.