AI Exoplanet Detection Algorithm
Develop a machine learning model to detect exoplanets from stellar light curves with >95% accuracy.
Problem Statement
Build an AI system to identify exoplanets using the transit method:
1. Dataset Analysis
- Kepler Space Telescope light curve data
- 100,000+ stellar observations
- Labeled true positives and false positives
2. Model Requirements
- Precision: >95%
- Recall: >90%
- Processing: <100ms per light curve
- Explain ability for scientific validation
3. Features to Detect
- Transit depth and duration
- Period consistency
- Secondary eclipses
- Multiple planet systems
4. Challenges
- Noise and stellar variability
- Rare positive examples (class imbalance)
- Different star types and sizes
- Instrumental artifacts
5. Innovation Areas
- Novel feature engineering
- Attention mechanisms
- Transfer learning approaches
- Real-time classification
Deliverables:
- Trained model with source code
- Performance metrics and validation
- Research paper draft
- API for integration