Training Data News and Research

RSS

AI Tracks Tropical Waves To Boost Hurricane Forecast Accuracy

Researchers at the University of Miami developed an AI tool that can automatically track tropical easterly waves (TEWs) and distinguish them from the Intertropical Convergence Zone and the monsoon trough. This innovation improves hurricane forecasting accuracy and provides forecasters with real-time tracking data.

21 Aug 2025

MIT Engineers Harness AI To Crack The Solubility Code For Drug Discovery

MIT chemical engineers have developed a machine learning model that predicts molecular solubility in organic solvents with high accuracy, streamlining pharmaceutical synthesis. The freely available tool could accelerate drug development and help identify safer, more sustainable solvents.

20 Aug 2025

New AI Method MIGHT Boosts Cancer Detection Accuracy While Reducing False Positives

Johns Hopkins researchers developed MIGHT and CoMIGHT, new AI methods that improve the reliability and accuracy of complex data analysis. Applied to liquid biopsy tests, they boosted early cancer detection while reducing false positives by accounting for inflammatory and vascular disease signals.

19 Aug 2025

Malicious Chatbots Trick Users Into Oversharing Personal Data, Study Finds

Malicious AI chatbots can manipulate users into revealing up to 12.5 times more personal information by mimicking empathy and emotional reciprocity, a new study by King’s College London reveals. The research highlights how easily large language models can be weaponized for covert data extraction, with users largely unaware of the privacy risks.

13 Aug 2025

Open Molecules 2025 Powers a New Era of AI-Driven Chemistry

Open Molecules 2025 is a landmark dataset of over 100 million quantum-level molecular simulations, developed by Meta and U.S. national labs, to train machine learning models for chemistry. It paves the way for rapid, low-cost predictions in drug discovery, energy storage, and materials science.

17 Jun 2025

New AI Method Cleans Dirty Data Before Training And Boosts Model Accuracy

Researchers at Florida Atlantic University have developed an automated method to detect and remove mislabeled data before training machine learning models, significantly improving their reliability and performance. The approach uses L1-norm principal component analysis to identify outliers without manual tuning or assumptions.

17 Jun 2025

Scientists Pinpoint When AI Turns From Helpful to Harmful

Researchers at George Washington University have derived an exact, reproducible formula that predicts when an AI language model's output will shift from accurate to misleading. The tipping point occurs when the model’s internal attention dynamics snap under pressure.

5 May 2025

PNNL Scientists Build Trustworthy AI That Knows Its Limits in Materials Discovery

Researchers at Pacific Northwest National Laboratory have developed a method to measure how well AI models, specifically neural network potentials, are trained to predict material properties. The approach quantifies uncertainty to enhance model trust and identifies areas where additional training is required.

25 Apr 2025

AI Predicts Kidney Cancer Drug Response Using Routine Pathology Slides

Researchers at UT Southwestern developed an AI model that uses histopathological slide images to predict which kidney cancer patients will benefit from anti-angiogenic therapy. This deep learning model rivals the accuracy of genetic tests like Angioscore but is more accessible and interpretable.

25 Apr 2025

FairAT Boosts AI Security and Fairness by Targeting Its Weakest Links

Researchers at Wuhan University developed Fair Adversarial Training (FairAT), a technique that enhances the fairness and robustness of AI models by focusing on their most vulnerable data points. FairAT significantly improves both equity and security, outperforming existing state-of-the-art methods.

18 Apr 2025

Physics Reveals How AI Attention Drives Repetition, Hallucination, and Bias—And How to Fix It

Researchers at George Washington University have developed a first-principles physics theory explaining how the attention mechanism in AI models like ChatGPT leads to repetition, hallucination, and bias.

16 Apr 2025

Digital Twin of Mouse Brain Predicts Neuron Activity and Connections With Striking Accuracy

Stanford-led researchers have developed a highly accurate AI-powered digital twin of the mouse visual cortex, trained on neural data from mice watching movies. This model can generalize responses to novel visual stimuli and predict neuron types and connections, revolutionizing brain experimentation.

10 Apr 2025

AI-Powered Brain Implant Restores Real-Time Speech for People With Paralysis

Researchers from UC Berkeley and UCSF have developed a brain-computer interface that decodes brain signals into audible speech in near real time. This AI-driven breakthrough enables more natural, fluent communication for individuals with severe paralysis.

1 Apr 2025

Wireless Networks Need Human-Like AI to Unlock the Next Tech Revolution

A groundbreaking paper argues that the next revolution in wireless technology hinges on AI with human-like common sense. The researchers propose a paradigm shift, integrating advanced AI with wireless networks to create an "AGI-native" system that can think, plan, and imagine like humans.

19 Mar 2025

AI Set to Transform Health Care as Experts Map Out Solutions to Overcome Barriers

AI has the potential to revolutionize health care by streamlining workflows, enhancing diagnostics, and improving patient outcomes, but barriers such as data privacy, high costs, and regulatory challenges slow its adoption. This comprehensive review offers a roadmap to overcome these hurdles and unlock AI's full potential in medicine.

16 Mar 2025

Generative AI Transforms Tunnel Safety With Smart Inspections and Autonomous Drones

Korea Institute of Civil Engineering and Building Technology has developed generative AI-powered inspection technology that creates synthetic images of tunnel damage to address data scarcity and enhance safety inspections. Integrated with autonomous drones, this innovation enables precise, efficient tunnel inspections without relying on large datasets.

12 Mar 2025

New AI Technique Cuts Hidden Biases Without Knowing What They Are

Researchers have developed a novel AI training technique that eliminates spurious correlations by pruning just a small subset of problematic data—without needing to know what those correlations are. This breakthrough enhances AI model reliability and performance across tasks.

11 Mar 2025