Aims and Scope
Generative AI (GenAI) is transforming the landscape of data mining and data science by enabling powerful data synthesis, feature creation, and knowledge discovery. As these models grow in scale and capability, they introduce new challenges related to data governance, privacy leakage, and traceability of generated outputs. A central tension emerges: data needed for effective model training remains distributed across organizations due to regulatory, security, and ethical constraints, while generative outputs themselves often lack verifiable provenance and transparency.
Federated Learning (FL) has become an important paradigm in data science for enabling distributed model training across siloed datasets. However, simply applying FL to GenAI does not resolve broader concerns around trust, content attribution, and algorithmic accountability. STFG aims to address these gaps by exploring "Trustworthy-by-Design" federated generative frameworks across the entire data lifecycle — from decentralized data collaboration (how models learn) to verifiable content generation and usage (how models produce and share knowledge).
Our workshop will provide a focused venue for the data mining community to examine the emerging Privacy–Utility–Provenance challenges in Federated GenAI. We aim to bridge advances in secure data mining, responsible AI, and distributed optimization to build generative systems that are private, auditable, and resilient to manipulation.
Significance
STFG contributes a novel perspective by integrating data mining, trustworthy AI, and federated model governance. Unlike traditional FL workshops that emphasize privacy-preserving training alone, STFG highlights the full generative workflow, including post-training risks such as data leakage through generated samples, unlearning in distributed systems, and provenance tracking.
As ICDM 2026 emphasizes Large Models and Ethical Data Analytics, this workshop directly addresses a key data science question: How can we ensure that generative models trained on decentralized, sensitive data remain safe, traceable, and compliant without centralized access to the underlying datasets? By bringing together researchers in privacy-preserving data mining, federated systems, generative modeling, and responsible AI, STFG aims to establish foundational methods and evaluation principles for secure and trustworthy GenAI in decentralized environments.
Topics of Interest
We welcome original research contributions on all aspects of secure and trustworthy federated generative AI. Topics include, but are not limited to:
Research Topics
- Private Collaborative Training
- Content Provenance & Accountability
- Reliable Deployment & Safety
- Knowledge-Enhanced Federated LLMs
- Evaluation & Benchmarking for Privacy
- Privacy in GenAI
- Automated Privacy Auditing for LLMs
- Ethical and Bias Alignment in FedGenAI
- Decentralized Model Personalization
Important Dates
- Paper submission deadline: June 6, 2026
- Author notification: August 16, 2026
- Camera-ready deadline: September 9, 2026
- Workshop day: November 12–15, 2026
- All times are 11:59 PM AoE (Anywhere on Earth)
Submission Guidelines
All papers must be submitted online via the wi-lab submission portal (link coming soon). Submitted manuscripts should use the ICDM 2026 paper template and be anonymized, i.e., papers must adhere to the IEEE 2-column format.
- For the regular paper track, submissions should not exceed 8 pages of content, plus an additional 2 pages for references.
- For the short paper track, submissions should be limited to a maximum of 4 pages of content, plus 1 extra page for references.
- In alignment with the ICDM 2026 reviewing scheme, all submissions will undergo triple-blind reviews by the Program Committee, evaluating technical quality, relevance to the workshop scope, originality, significance, and clarity.
- Duplicate submissions of the same paper are forbidden (including to more than one ICDM workshop). Papers must be original research. Submitting a paper substantially similar in content to a paper that has been accepted or is under consideration at another archival venue is not allowed. During the review process, or after acceptance, submitted papers cannot be submitted to another archival venue unless substantial new material is added.
- Accepted papers will be included in the ICDM Workshop Proceedings (separate from ICDM Main Conference Proceedings). The proceedings are published by IEEE and EI-indexed. Each accepted workshop paper requires a full registration.
- A paid registration is required for each accepted paper, regardless of whether it is presented in person or via video.
- Desk Reject Policy: Submissions that fail to adhere to the anonymity, length, or formatting requirements, or are affected by academic dishonesty issues, such as plagiarism, author misrepresentation, or falsification, may be subject to desk rejection by the chairs.
Presentation Guidelines: Accepted papers will be presented as posters or pre-recorded videos. Selected papers will be invited for spotlight oral presentations.
Program Committee
Expert in Data Security and Privacy, Recommender Systems, and Federated Learning
huan.huo@uts.edu.au
Expert in AI, LLM, Causal Learning, and Federated Learning
dianer.yu1@unsw.edu.au
Expert in AI Security, Machine Learning, and Federated Learning
zhwu@sem.ecnu.edu.cn
Expert in Federated Learning, Data Analytics, and AI Applications
duhaizhou@shiep.edu.cn
Expert in Deep Learning, Recommender Systems, and Federated Learning
jlzhao@sdust.edu.cn
Expert in Generative AI, LLM, and Federated Learning
ling.chen@uts.edu.au
Contact
For queries regarding STFG @ ICDM 2026, please contact the organizing committee or visit the ICDM 2026 official website.