Science

Researchers have developed a powerful framework for accurately reconstructing complex network structures from observational data, even in directed graphs.

New research explores how connecting large language models with focused knowledge bases improves their ability to answer complex questions about diseases like Alzheimer’s and Type 2 Diabetes.

A new approach simplifies the core architecture of the Decision Transformer by eliminating unnecessary historical data, leading to faster and more efficient offline reinforcement learning.

A new approach combines the power of machine learning with large-scale optimization to deliver personalized recommendations that satisfy multiple stakeholders and complex constraints.

Researchers demonstrate that providing artificial intelligence with structural guidance dramatically improves its ability to automatically prove complex mathematical theorems, even with limited computing power.

A new deep-learning approach, combined with advanced holographic microscopy, is providing unprecedented insight into the identification and characterization of microscopic plastic fibers in our environment.

New research uses economic principles and a novel simulator to understand why short-form video is so addictive, and how platforms can design more responsible recommendation systems.

A new analysis reveals how contaminated training sets-even those incorporating realistic synthetic data-can significantly degrade the performance of common learning algorithms.

A new method dramatically speeds up the process of identifying vulnerabilities in large language models, offering a more practical approach to AI security.

Researchers have developed a novel diffusion model that tackles the challenge of missing data in sequential recommendation systems, improving accuracy and personalization.