🤖
Hard
AI & ML

AI Exoplanet Detection Algorithm

Develop a machine learning model to detect exoplanets from stellar light curves with >95% accuracy.

Prize Pool
$30,000
Participants
890
Deadline
2/15/2025
Time Left
-362 days

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

Required Skills

PythonTensorFlow/PyTorchSignal ProcessingAstronomy

Mentors

Dr. Lisa Wang - SETI
Alex Kumar - SpaceX ML