IJCNN25-SS-ResponsibleFM

Background

With the fast development of machine learning algorithms, foundation models (FMs) have been proposed to address multiple tasks at the same time. A well-known example is ChatGPT which has significantly impacted various aspects of our daily lives. Recent achievements of FMs have positioned neural network architectures as both a powerful solution and one of the most promising methodologies for advancing toward artificial general intelligence (AGI). Although FMs have shown great abilities in solving many problems, they still face significant challenges concerning reliability when deploying them in real-world environments.

Aims

As the potential applications of FMs continue to expand, it is urgent to push studies regarding trusted and responsible FMs. The powerful capabilities and broad applicability of FMs highlight the need for responsible design and usage, emphasizing the importance of enhanced interpretability, trustworthiness, and new learning paradigms to meet these evolving requirements. Addressing these challenges requires a deeper understanding of both the theoretical foundations and the evolving methodologies or applications of FMs. This special session aims to foster the exchange of ideas in trustworthy machine learning and responsible AI in the wild, encompassing new theories, innovative algorithms, and their practical applications.

Main Topics

We welcome submissions in but not limited to the following aspects.

Reliable training (e.g., pretraining and fine-tuning) and inference methods for FMs
Theory, methodology, and applications of handling out-of-distribution data in the era of FMs, and the hallucination detection and mitigation of LLM and MLLM
Theory and applications of safety research for FMs, such as jailbreaking, red teaming, and watermarking issues
Theory and applications of identification of machine-generated contexts by FMs
Theory and applications of the deployment of FMs with safety, ethics, and fairness
Theory and applications of adaptation of FMs to particular tasks
Theory, methodology, and application of FM pre-training or fine-tuning with incremental tasks or instructions in real world, e.g., continual learning, unlearning, model editing
Theory and methodology for enhancing the responsibility, trustworthy, and interpretability of FMs, such as causal machine learning, learning with uncertainty, etc
Theory, methodology, and application of FM-based prompt engineering and prompt learning
Privacy-leakage evaluation of FM (e.g., membership inference attack, model inversion attack)
Evaluation of the reliability of foundation models, e.g., the benchmark dataset, platform or measure to evaluate foundation models in real-world contexts
Applications of FMs for scientific domains, including program languages, climate science, healthcare, life sciences, physics, and cognitive science

Submissions

Please refer to https://2025.ijcnn.org/authors/initial-author-instructions for more details.

Format

Official IEEE double-column format download from Download the Paper Template

Pages

6-8 pages. A maximum of two extra pages per paper is allowed (i.e, up to 10 pages) with additional charge.

Double-Blind Reviewing

The author and affiliation information are not shown in any part of the manuscript.

Submission Link

Submit your paper through Microsfot CMT

Submission Notes

Choose "Responsible Foundation Models in the Wild" as your primary.

Registration

Each paper needs to be covered by at least one registration if accepted.

Organisers

Yiliao Song
(The University of Adelaide, AU)

Xuefeng Du
(University of Wisconsin-Madison, US)

Zijian Wang
(The University of Queensland, AU)

Zhen Fang
(University of Technology Sydney, AU)

Yadan Luo
(The University of Queensland, AU)

Feng Liu
(The University of Melbourne, AU)

Dong Gong
(The University of New South Wales , AU)

Sharon Yixuan Li
(University of Wisconsin-Madison, US)

Paper Submission Deadline	January 31, 2025
Notification of Acceptance	March 31, 2025
Camera-Ready Paper Submission	May 1, 2025
Early Registration Deadline	May 1, 2025
Conference Dates	June 30 – July 5, 2025

IJCNN 2025 Special Session on

Responsible Foundation Models in the Wild

Background

Aims

Main Topics