About
Research & Development
Pinpointing Why AI Struggles with Data Visualizations
TLDR: Current vision-language models (VLMs) struggle with understanding data visualizations. Researchers developed FUGU, a new benchmark, and used interpretability techniques to diagnose these failures....
Research & Development
Enhancing Artistic Control in AI Image Generation with Proportion and Perspective ControlNets
TLDR: Researchers introduced two new ControlNet modules for Flow Matching-based diffusion models: a Proportion ControlNet using bounding boxes for object placement and scale, and...
Research & Development
Predicting and Preventing Post-Surgery Aspiration: Insights into Opioids, Surgical Sites, and Gender Differences
TLDR: A new study developed a machine learning model to predict postoperative aspiration, achieving an AUROC of 0.86. Key predictors include maximum daily opioid...
Research & Development
Advancing Collective AI Decisions with Online Mixture of Experts
TLDR: This research introduces Online Mixture of Experts (OMoE), a new framework for optimal collective decision-making using expert-guided bandit learning. It proposes two algorithms:...
Research & Development
Advancing Madden-Julian Oscillation Forecasts with a New AI Correction System
TLDR: A new AI model, PCC-MJO, significantly improves Madden-Julian Oscillation (MJO) predictions by acting as a post-processor for existing dynamical models. It extends skillful...
Research & Development
DiffGRM: A Diffusion Model for Enhanced Generative Recommendation
TLDR: DiffGRM is a new generative recommendation model that uses a masked discrete diffusion model to overcome limitations of traditional autoregressive models. It introduces...