Milad Abdollahzadeh*, Guimeng Liu*, Touba Malekzadeh†, Christopher T. H. Teo†, Keshigeyan Chandrasegaran†, Ngai-Man Cheung‡
(* Equal first-author contribution, † Equal second-author contribution，‡ Corresponding author)

This repo contains the list of papers with public code implementations for Generative Modeling under Data Constraint (GM-DC). For each work, we determine the generative task(s) addressed, the approach, and the type of generative model used.

First, we define the generative tasks, and the approach definition, and then provide our comprehensive list of the works for GM-DC with the required details for each work.

• Dec 24, 2025: Our paper is accepted by TMLR!!! 🎉 Awarded Survey Certificate for "an exceptionally thorough or insightful survey".
• July 15, 2025: 120 new works added (233 works total)!
• Oct 28, 2024: The slides for our ICIP tutorial on "Generative Modeling for Limited Data, Few Shots and Zero Shot" can be found here.
• July 28, 2023: First release (113 works included)!

In machine learning, generative modeling aims to learn to generate new data statistically similar to the training data distribution. In this paper, we survey learning generative models under limited data, few shots and zero shot, referred to as Generative Modeling under Data Constraint (GM-DC). This is an important topic when data acquisition is challenging, e.g. healthcare applications. We discuss background, challenges, and propose two taxonomies: one on GM-DC tasks and another on GM-DC approaches. Importantly, we study interactions between different GM-DC tasks and approaches. Furthermore, we highlight research gaps, research trends, and potential avenues for future exploration.

We define 8 different generative tasks under data constraints based on the rigorous review of the literature. The description of these tasks can be found in the follwing table:

Task	Description & Example	Illustration
uGM-1	Description: Given $K$ samples from a domain $\mathcal{D}$, learn to generate diverse and high-quality samples from $\mathcal{D}$ Example: ADA learns a StyleGAN2 using 1k images from AFHQ-Dog
uGM-2	Description: Given a pre-trained generator on a source domain $\mathcal{D}_s$ and $K$ samples from a target domain $\mathcal{D}_t$, learn to generate diverse and high-quality samples from $\mathcal{D}_t$ Example: CDC adapts a pre-trained GAN on FFHQ (Human Faces) to Sketches using 10 samples
uGM-3	Description: Given a pre-trained generator on a source domain $\mathcal{D}_s$ and a text prompt describing a target domain $\mathcal{D}_t$, learn to generate diverse and high-quality samples from $\mathcal{D}_t$ Example: StyleGAN-NADA adapts pre-trained GAN on FFHQ to the painting domain using Fernando Botero Painting as input
cGM-1	Description: Given $K$ samples with class labels from a domain $\mathcal{D}$, learn to generate diverse and high-quality samples conditioning on the class labels from $\mathcal{D}$ Example: CbC trains conditional generator on 20 classes of ImageNet Carnivores using 100 images per class
cGM-2	Description: Given a pre-trained generator on the seen classes $C_{seen}$ of a domain $\mathcal{D}$ and $K$ samples with class labels from unseen classes $C_{unseen}$ of $\mathcal{D}$, learn to generate diverse and high-quality samples conditioning on the class labels for $C_{unseen}$ from $\mathcal{D}$ Example: LoFGAN learns from 85 classes of Flowers to generate images for an unseen class with only 3 samples
cGM-3	Description: Given a pre-trained generator on a source domain $\mathcal{D}_s$ and $K$ samples with class labels from a target domain $\mathcal{D}_t$ , learn to generate diverse and high-quality samples conditioning on the class labels from $\mathcal{D}_t$ Example: VPT adapts a pre-trained conditional generator on ImageNet to Places365 with 500 images per class
IGM	Description: Given $K$ samples (usually $K=1$) and assuming rich internal distribution for patches within these samples, learn to generate diverse and high-quality samples with the same internal patch distribution Example: SinDDM trains a generator using a single image of Marina Bay Sands, and generates variants of it
SGM	Description: Given a pre-trained generator, $K$ samples of a particular subject, and a text prompt, learn to generate diverse and high-quality samples containing the same subject Example: DreamBooth trains a generator using 4 images of a particular backpack and adapts it with a text-prompt to be in the grand canyon

Please refer to our survey for a more detailed discussion of these generative tasks including the attributes of each task and the data limitation range that addressed for each task.

If you find this repo useful, please cite our paper

@article{abdollahzadeh2025a,
  author    = {Milad Abdollahzadeh 
              and Guimeng Liu 
              and Touba Malekzadeh 
              and Christopher T.H Teo 
              and Keshigeyan Chandrasegaran 
              and Ngai-Man Cheung
                  },
  title     = {A Survey on Generative Modeling with Limited Data, Few Shots, and Zero Shot},
  journal   = {Transactions on Machine Learning Research},
  issn={2835-8856},
  year={2025},
  url={https://openreview.net/forum?id=u7GTHazuRp},

}