Researchers Have Found a New Way to Enhance AI Performance Across Image Editing, Drug Discovery

Researchers at Clarkson University, US, have developed a new mathematical framework that could make artificial intelligence systems more accurate and controllable across applications ranging from image editing to drug discovery and scientific simulations.

The framework, called Rex, was created by Zander Blasingame, a postdoctoral researcher, and Chen Liu, Professor of Electrical and Computer Engineering at Clarkson University. Their research paper, “Rex: A Family of Reversible Exponential (Stochastic) Runge-Kutta Solvers,” has been selected for presentation at the International Conference on Machine Learning (ICML) 2026, one of the world’s leading AI conferences.

The new framework focuses on improving diffusion and flow-matching models, which form the foundation of many modern generative AI systems, including image generators, molecular design tools, and scientific simulators. These models typically work by transforming random noise into useful outputs such as images or molecular structures.

While this process is effective for generating new content, many applications require AI systems to reverse it and recover the original information. Existing methods often introduce errors during this step, limiting accuracy and control.

According to the researchers, Rex addresses this challenge by aligning the forward and reverse processes more closely, enabling AI systems to retrace their steps with significantly greater precision.

“One important application of exact inversion solvers is round-trip image editing. While all reversible methods accumulate some numerical error, Rex achieves orders-of-magnitude lower inversion error than competing approaches,” Blasingame said in a statement.

The researchers said the framework could lead to more precise image-editing tools by allowing users to modify AI-generated images while preserving important details. Beyond creative applications, the technology could also improve molecular simulations used in chemistry and drug discovery, where accurately modelling complex processes is critical.

A key advantage of the framework is that it can be integrated into existing diffusion and flow-matching models without requiring developers to redesign entire AI systems. This could make adoption easier for researchers and companies already working with generative AI technologies.