Trustworthiness in LLMs

Large Language Models (LLMs) have transformed natural language processing, demonstrating remarkable capabilities across diverse tasks. As these models become increasingly integrated into real-world applications, ensuring their responsible and trustworthy behaviors is crucial for their broader adoption and effective deployment. This special track will provide a leading forum for researchers and practitioners to share cutting-edge techniques designed to enhance the trustworthiness of LLMs. Participants will gain a comprehensive understanding of methods and tools for designing, evaluating, and deploying LLMs with a focus on privacy, security, robustness, and overall trustworthiness.

We encourage state-of-the-art research on LLMs, focusing on privacy, security, robustness, transparency, and trustworthiness. In addition, we encourage studies that explore the application of LLMs across diverse domains, such as natural language processing, computer vision, healthcare, sensor data applications, environmental science, etc. This track aims to bring researchers and practitioners from artificial intelligence, information systems, privacy and security, and various application domains to advance the frontiers of frameworks, theories, and methodologies.

Topics of interest include, but are not limited to:

  • Novel methods for building trustworthy LLMs
  • New applications and settings where the trustworthiness of LLMs plays an important role and how well existing techniques work under these settings
  • LLMs with verifiable guarantees (such as robustness, fairness, and privacy guarantees) to build trustworthiness
  • Privacy-preserving LLM approaches
  • Theoretical understanding of trustworthy LLMs
  • Explainable and interpretable LLMs
  • Robust decision-making under uncertainty in LLMs
  • Bias and fairness concerns about trustworthy LLMs
  • Case studies and field research on the societal impacts of applying LLMs

Accepted papers will be published in the conference proceedings; also, extended versions of selected best papers will be invited for publication in Journal of Combinatorial Optimization, and Applied Network Science.

Track Co-Chairs: Phung Lai (SUNY-Albany) and Xin Li (SUNY-Albany)

We follow the manuscript submission, review methods and deadlines, set by CSoNet 2024.