Classify your Transit data through our state of the art ML Algorithm or create your own classification algorithm
Classify your newly acquired data with our collection of pre-trained models trained on real NASA data. Get instant results with state-of-the-art machine learning algorithms.
Choose from CNN, DNN, or Classical models
Transit light curves or time-series data
Real-time classification with confidence scores
Download results or use our API
New to Machine Learning? Create your own machine learning algorithm to classify real world Transit data. Learn by doing with our interactive tutorials and guided projects.
Pre-processed NASA data or upload your own
Random forests, neural networks, and more
Watch your model learn with real-time metrics
Evaluate performance and export your model
Build and train models with our intuitive interface. Real-time feedback helps you understand each step.
Watch your models learn in real-time with live visualization of training metrics and performance indicators.
Download your trained models in multiple formats including TensorFlow, PyTorch, and ONNX for deployment.
Explore your datasets with interactive charts, plots, and 3D visualizations to gain deeper insights.
Upload your own astronomical data or use our curated collection of NASA exoplanet transit datasets.
Access comprehensive tutorials, documentation, and guided learning paths for all skill levels.
Our platform combines cutting-edge machine learning with intuitive design to make exoplanet classification accessible to everyone.
Upload your transit light curve data or select from our curated NASA dataset collection with over 100,000 confirmed signals.
Choose from pre-trained models or build your own using our drag-and-drop interface with various ML algorithms.
Watch as your model learns patterns and provides real-time insights into exoplanet signals with live visualization.
Get detailed classification results with confidence scores and export for further research or integration.
< 4 seconds per light curve classification
more than 90% accuracy on NASA validation dataset
Supports multiple data input methods and formats
Full Stack Developer
Machine Learning Engineer
NASA
Space Apps Participant
Data Scientist
Exoplanet Researcher
ML Algorithm
Specialist
Everything you need to know about Exchron and exoplanet classification
Exchron offers two distinct modes: Playground Mode for quick experimentation with pre-trained exoplanet classification models, and Classroom Mode for building and training custom models from scratch. Playground is perfect for immediate results and testing, while Classroom provides a comprehensive learning environment with guided tutorials.
Playground Mode follows a 4-step workflow: 01 Overview (explore and select pre-trained models), 02 Data Input (choose from preloaded data, manual entry, or file upload), 03 Results (view predictions and performance metrics), and 04 Enhance (improve model performance and provide feedback).
You can input data through four methods: Preloaded Data (curated Kepler and TESS datasets), Manual Entry (interactive parameter sliders), Data Upload (CSV files up to 5MB), and Batch Processing (coming soon). We support Kepler Objects of Interest (4,892 samples) and TESS Objects of Interest (2,674 samples).
Classroom Mode provides a complete machine learning pipeline: Data Input (upload CSV files with automatic analysis), Model Selection (choose Neural Networks or Random Forest with hyperparameter tuning), Train & Validate (real-time training monitoring), and Test & Export (performance evaluation and model download).
Exchron requires a modern browser with JavaScript enabled and 4GB+ RAM for optimal performance. Data uploads are limited to 10MB CSV files with up to 100,000 rows and 300 columns. All computation runs client-side, so no server connection is required after initial loading.
Models can be exported in multiple formats including TensorFlow.js and JSON. Export options include trained model files, metadata with training configuration, comprehensive performance reports with visualizations, and ready-to-use inference code snippets for deployment.
Still have questions?
Contact Support