Despite the substantial benefits from using synthetic data, the process of synthetic data generation is still an ongoing technical challenge. Although the two scenarios of limited data and privacy concerns share similar technical challenges such as quality and fairness, they are often studied separately. We invite researchers to submit papers that discuss challenges and advances in synthetic data generation, including but not limited to the following topics.
- Methods for generating synthetic data including but not only limited to:
o ML and DL generative techniques for various modalities of data
o LLM-adaptation in multi-tabular data synthesizer framework
o Data pre-processing to leverage the strength of LLMs
o Missing data handling using LLM
- LLM-synthesized data quality evaluation, including but not only limited to:
o Fidelity metrics of LLM-generated data compared to traditional synthesizers
o Synthetic data utility level in downstream tasks between LLM synthesizers and traditional synthesizers
o LLM adaptation in synthetic data evaluation metrics
- Societal and practical use cases, including but not only limited to:
o Privacy and fairness issues that may arise in certain financial tasks
o Justification of GenAI model adaptations in financial practice
o Legal hurdles and implications of synthetic data usage in finance
IMPORTANT DATES:
Paper submission: October 14, 2024 (AOE) Oct 21, 2024 (AOE)
Acceptance Notification: October 25, 2024
Workshop: Nov 14, 2024
Submission Process
In addition to the paper content (PDF document), the paper title, author names, contact details, and the text, a brief abstract must be submitted electronically through the workshop's submission site.
** At least one author of each accepted paper is required to attend the conference to present the work **
Review Process
This workshop will follow a single-blind review process, meaning that the authors’ identities will be known to the reviewers, but the reviewers’ identities will not be known to the authors.
There will be no rebuttal period.
Policy on multiple submissions and pre-publication
Please do not submit any paper that at the time of submission is already published or accepted for publication in a journal or other archival publication. Questions about submission eligibility should be referred to the program chairs in advance of the submission deadline.
Submitted papers may have appeared previously as a technical report or in other non-archival venues. Specific examples of permissible venues include ICML workshops, NeurIPS workshops, AAAI workshops, and arXiv. Questions concerning acceptable pre-publication venues should be directed to the program chairs.
All submissions will be treated with strict confidentiality until published.
The workshop is non-archival and will not have official proceedings. However, accepted papers will be linked on our website.