SyntheticData4ML Workshop

We are pleased to announce that we will host the SyntheticData4MLWorkshop as part of NeurIPS 2022 on 2 December 2022. The workshop brings together research communities in generative models, privacy, and fairness as well as industry leaders in a joint effort to develop the theory, methodology, and algorithms to generate synthetic benchmark datasets with the goal of enabling ethical and reproducible ML research.

We invite submissions of 4 – 6 pages (references not included) of original papers on the methodology and applications of synthetic data. Example topics include (but are not limited to):
 
New machine models for generating synthetic data of various modalities (tabular data, image data, time series data, high-dimensional, etc.)
Privacy-preserving synthetic data (e.g. theory, generation algorithm, attack modes)
Cross-domain data generation (e.g. generation of out-of-distribution data)
Causal generative models
Synthetic data and data augmentation for ML fairness
Synthetic data for specific domains and tasks (e.g. architectures, inductive biases)
Synthetic data evaluation and performance metrics (e.g. fidelity, utility, privacy)
Novel applications of synthetic data in various application domains (medical, pharmaceutical, financial, education, etc.)
Comparison to alternative solutions for secure/private data exchange
 
Papers may include unlimited supplementary material, but reviewers will not be obliged to read these supplementary materials. The authors should use the NeurIPS style file provided here. All submissions are expected to conform to the NeurIPS code of conduct.