Machine learning

A new approach to training AI models can help improve fairness in decision outcomes for the most disadvantaged groups even when their identities are unknown.

Stress-testing artificial intelligence models under real-world conditions can improve their reliability, enabling the maritime industry to embark on safer and greener voyages.

Inspired by classical signal processing techniques, a new approach to fine-tuning AI models helps reduce irrelevant computing while maintaining performance.

A new attribute-based approach to object detection helps visual artificial intelligence models spot unknown objects more accurately and efficiently.

By adding rewards for logical reasoning steps, a new training framework aims to make large language models more robust and reliable.

A machine learning model called Fragle helps detect cancer and spot signs of relapse by quantifying DNA fragments in the blood.

A consistent rulebook of prompts could be the key to advancing artificial intelligence models, enhancing their image classification performance even when dealing with unknown objects.

A new framework aligns data from multiple sensors across time, boosting the accuracy of machine failure predictions.

A new AI-driven computational pipeline paired with rigorous lab-based testing accelerates the discovery of novel molecules for efficient organic solar cells.

A new machine learning tool could help predict cancer treatment outcomes for more effective and targeted treatments.

Departing from costly simulations and trial-and-error methods, researchers harness machine learning and automation to efficiently discover new recipes for green hydrogen catalysts.

A new machine learning model identifies key dementia features from complex clinical datasets, potentially making diagnoses faster and less subjective.