Skip to main content

Model Training

Overview​

Model training is designed to be simple yet powerful. With just a few clicks, you can train sophisticated machine learning models on your hyperspectral data. Clarity handles the complexity behind the scenes, allowing you to focus on your analysis rather than technical details.

Creating a model​

To create a new model, go to the Models page and click New model.

The Training Process​

Step 1: Model Configuration​

Training is as easy as selecting three key components:

The default parameters are optimized to work for most use cases, so you can get started immediately without worrying about complex configuration.

Step 2: Start Training​

Click the Train button to begin the training process. Clarity will automatically:

  • Validate your dataset compatibility
  • Set up the optimal training environment
  • Begin training with the selected parameters
  • Monitor progress and performance

Step 3: Automatic Testing​

If your dataset includes test data, Clarity will automatically:

  • Perform testing once training is complete
  • Generate inference results
  • Calculate performance metrics

What You Need​

Essential Requirements​

  • Dataset: Must be finalized and contain properly labeled data
  • Model Type: Choose based on your analysis goals (see Model Types for details)
  • Training Data: Sufficient samples for each class with clear spectral distinctions

For Regression Models​

  • Label Attributes: Regression models require labels to have numerical attributes
  • Continuous Values: Ensure your labels represent the continuous variables you want to predict
  • Data Distribution: Training data should cover the range of values you expect to predict

Advanced Settings​

If you need to customize your model training, advanced settings are available. These include model architecture selection, learning rate optimization, batch size configuration, and early stopping parameters.

Learn more about Advanced Settings →

Next steps​