The ISIT 2024 Workshop on Information-Theoretic Methods for Trustworthy Machine Learning (IT-TML) aims to establish a platform where researchers and engineers can come together to address the challenges and propose solutions pertaining to the responsible deployment of ML in applications of social consequence, with a particular focus on privacy and fairness. 

Recent developments in privacy, fairness, and robustness underscore the crucial role that information theory is poised to play in the upcoming decade of machine learning (ML) applications. Information-theoretic approaches are instrumental in refining generalization bounds for deep learning, offering robust assurances for compressing neural networks, fostering fairness and privacy in both ML training and deployment, enhancing communication efficiency in distributed training, and elucidating the constraints associated with learning from noisy data. This workshop aims to spotlight these emerging and socially significant research domains, providing ISIT attendees with insights into the information-theoretic methodologies underpinning these recent advancements.

Accepted papers will be presented as posters during the poster session, and will not be included in any formal proceedings. As such, we welcome submissions that have been presented or published elsewhere. The titles of accepted papers will be featured on the workshop website, accompanied by a link to the full version that authors will publicly share by the camera-ready deadline.  Some papers will also be selected for spotlight presentations.

The deadline is March 3

Read the call for papers