Project Overview
Siamese Neural Network System for Exoplanet Transit Detection
Last Model Sync: 12m ago
Model Accuracy
98.4%
+0.2% vs previous build
Confirmed Planets
4,152
Across K2/TESS datasets
False Positive Rate
1.2%
Optimized threshold
Processed Flux Points
12.4M
Real-time telemetry
System Workflow
RAW
Kepler Flux
SNN
Siamese Model
PRED
Probability
Our system utilizes a Siamese Twin Architecture to compare unknown stellar flux patterns against known transit signatures, enabling high-precision detection of periodic brightness dips.
Training Loss
Epochs (1-200)
V2.4 RELEASED
Model Performance Metrics
F1-Score: 0.978 | Precision: 0.982 | Recall: 0.974 | Matthews Correlation Coefficient: 0.965