Powerful AI finds 100+ hidden planets in NASA data including rare and extreme worlds

2026-05-03 · Source: Robotics Research News -- ScienceDaily · Field: Science & Research — Artificial Intelligence & Machine Learning, Space Science & Astronomy, Research Methodology & Innovation · Depth: Expert, short

Summary

Astronomers at the University of Warwick have confirmed over 100 exoplanets, including 31 new discoveries, by applying a novel artificial intelligence system named RAVEN to data from NASA's Transiting Exoplanet Survey Satellite (TESS). This system analyzed observations from more than 2.2 million stars over TESS's first four years, focusing on planets with orbital periods under 16 days. The findings, published in *MNRAS*, include ultra-short-period planets, worlds within the "Neptunian desert," and tightly packed multi-planet systems. RAVEN's strength lies in its ability to process the entire detection workflow, from signal identification to statistical validation, using machine learning models trained on realistically simulated astrophysical events. This has led to one of the most precise measurements of close-in planet prevalence, finding that 9-10% of Sun-like stars host such planets, and directly measuring the rarity of Neptunian desert planets at 0.08% of Sun-like stars.

Key takeaway

For AI Scientists developing astronomical data analysis tools, RAVEN demonstrates the power of end-to-end automated pipelines in transforming raw telescope data into validated discoveries. You should consider integrating machine learning models trained on extensive, realistically simulated datasets to improve signal vetting and reduce false positives. This approach not only accelerates discovery but also yields cleaner datasets for robust population studies, enabling more precise measurements of planetary prevalence and characteristics.

Key insights

AI-driven pipelines can significantly enhance exoplanet detection and population studies by filtering false positives and correcting observational biases.

Principles

• Simulated data improves AI model training for astrophysical event classification.
• Automated pipelines enable consistent, objective analysis of large datasets.
• Correcting for detection biases yields more reliable population statistics.

Method

RAVEN detects stellar dimming, vets signals using machine learning trained on simulated planets and false positives, and statistically validates candidates, handling the entire process from detection to validation.

In practice

• Utilize AI for large-scale astrophysical data analysis.
• Develop comprehensive simulated datasets for model training.
• Release interactive catalogs for community follow-up observations.

Topics

• RAVEN Pipeline
• Exoplanet Confirmation
• TESS Mission
• Ultra-Short-Period Planets
• Neptunian Desert

Best for: AI Scientist, Research Scientist

Related on AIssential

• Making Sense of the Early Universe — AI and GPU acceleration dramatically enhance astronomical discovery by automating complex data analysis.
• TransitNet: A Compact Attention-Augmented Deep Learning Framework for Low-SNR Transit Blind Searches — TransitNet significantly improves low-SNR exoplanet transit detection using a compact, attention-augmented deep learning approach.
• Using Asta AutoDiscovery: AI-powered autonomous scientific discovery — AutoDiscovery uses Bayesian surprise to autonomously find unexpected patterns and insights in data.
• FLAME 2023: Diving into the Future of Fluid Dynamics and Machine Learning — AI techniques are transforming simulation by integrating physics-based modeling with data-driven machine learning.
• Gemini for Science: AI experiments and tools for a new era of discovery — Gemini for Science leverages AI agents to automate complex research tasks, significantly accelerating scientific discovery and exploration.
• Andrej Karpathy Open-Sources ‘Autoresearch’: A 630-Line Python Tool Letting AI Agents Run Autonomous ML Experiments on Single GPUs — Autoresearch enables AI agents to autonomously optimize ML models on single GPUs through iterative, metric-driven refinement.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Robotics Research News -- ScienceDaily.