TorchSOM Logo

Getting Started

  • Installation
    • Requirements
      • Systems
      • GPU
    • Installation Methods
      • Install from PyPI (Recommended)
      • Install from Source
      • Ensure GPU Support
    • Verification
    • Dependencies
      • Core Dependencies
      • Optional Dependencies
    • Getting Help
    • Next Steps
  • Quick Start Guide
    • Your First SOM
    • Basic Visualization
    • Common Parameters
    • Complete Example
    • What’s Next?
  • Basic Concepts
    • What is a Self-Organizing Map?
    • How SOMs Work
      • The SOM Algorithm
      • Mathematical Foundation
    • Core Components
      • 1. Grid Topology
      • 2. Neighborhood Function
      • 3. Learning Rate Decay
      • 4. Distance Functions
      • 5. Quality Metrics
    • Strengths and Weaknesses
      • Advantages
      • Limitations
    • Best Practices
      • Data Preparation
      • Parameter Selection
      • Interpretation
    • Next Steps

User Guide

  • SOM Visualization Guide
    • Overview
    • Quick Start
      • Basic Visualization Setup
      • Custom Configuration
    • Visualization Types
      • Training Errors
      • Distance Map (D-Matrix)
      • Hit Map
      • Component Planes
      • Supervised Maps
        • Classification Case
        • Regression Case
      • Advanced Visualizations
        • Score Map
        • Rank Map
    • Advanced Usage Examples
      • Batch Visualization Generation
      • Custom Colormap Usage
      • Troubleshooting
    • References

API Reference

  • Core API
    • Base Classes
      • BaseSOM
        • BaseSOM.build_distance_map()
        • BaseSOM.build_hit_map()
        • BaseSOM.fit()
        • BaseSOM.identify_bmus()
        • BaseSOM.initialize_weights()
        • BaseSOM.quantization_error()
        • BaseSOM.topographic_error()
    • Classical SOM Implementation
      • SOM
        • SOM.build_bmus_data_map()
        • SOM.build_classification_map()
        • SOM.build_distance_map()
        • SOM.build_hit_map()
        • SOM.build_metric_map()
        • SOM.build_rank_map()
        • SOM.build_score_map()
        • SOM.collect_samples()
        • SOM.fit()
        • SOM.identify_bmus()
        • SOM.initialize_weights()
        • SOM.quantization_error()
        • SOM.topographic_error()
    • SOM Variants
      • Growing SOM
      • Hierarchical SOM
  • Utilities API
    • Distance Functions
    • Neighborhood Functions
    • Decay Functions
    • Grid and Topology
      • adjust_meshgrid_topology()
      • axial_distance()
      • convert_to_axial_coords()
      • create_mesh_grid()
      • offset_to_axial_coords()
      • get_all_neighbors_up_to_order()
      • get_hexagonal_offsets()
      • get_rectangular_offsets()
    • Initialization
      • initialize_weights()
      • pca_init()
      • random_init()
    • Metrics
      • calculate_quantization_error()
      • calculate_topographic_error()
  • Visualization API
    • SOM Visualizer
      • SOMVisualizer
        • SOMVisualizer.plot_all()
        • SOMVisualizer.plot_classification_map()
        • SOMVisualizer.plot_component_planes()
        • SOMVisualizer.plot_distance_map()
        • SOMVisualizer.plot_grid()
        • SOMVisualizer.plot_hit_map()
        • SOMVisualizer.plot_metric_map()
        • SOMVisualizer.plot_rank_map()
        • SOMVisualizer.plot_score_map()
        • SOMVisualizer.plot_training_errors()
    • Visualization Configuration
      • VisualizationConfig
        • VisualizationConfig.cmap
        • VisualizationConfig.colorbar_pad
        • VisualizationConfig.dpi
        • VisualizationConfig.figsize
        • VisualizationConfig.fontsize
        • VisualizationConfig.fontweight
        • VisualizationConfig.grid_alpha
        • VisualizationConfig.hexgrid_size
        • VisualizationConfig.save_format
    • Supported Formats
  • Configuration API
    • SOM Configuration
      • SOMConfig
        • SOMConfig.batch_size
        • SOMConfig.device
        • SOMConfig.distance_function
        • SOMConfig.epochs
        • SOMConfig.initialization_mode
        • SOMConfig.learning_rate
        • SOMConfig.lr_decay_function
        • SOMConfig.model_config
        • SOMConfig.neighborhood_function
        • SOMConfig.neighborhood_order
        • SOMConfig.random_seed
        • SOMConfig.sigma
        • SOMConfig.sigma_decay_function
        • SOMConfig.topology
        • SOMConfig.x
        • SOMConfig.y
    • Saving Configuration
      • Loading Configuration
      • Exporting Configuration

Additional Resources

  • Changelog
    • How to Contribute
      • Report Issues
  • Frequently Asked Questions
    • General Questions
      • What is TorchSOM?
      • How does TorchSOM differ from other SOM implementations?
    • Installation and Setup
      • Which Python versions are supported?
      • Do I need a GPU to use TorchSOM?
    • Data Preprocessing
      • Should I always normalize my data?
      • What about categorical features?
    • Performance and Optimization
      • My training is very slow. How can I speed it up?
      • How much memory does TorchSOM use?
    • Visualization Issues
      • Why are some neurons white in my visualizations?
      • How do I interpret the distance map (D-Matrix)?
      • Can I customize the visualization colors?
    • Advanced Topics
      • Can I use TorchSOM for time series data?
      • How do I implement custom distance functions?
      • Can I save and load trained SOMs?
    • Integration Questions
      • How do I cite TorchSOM in my research?
    • Getting Help
      • Where can I get more help?
      • How do I report a bug?
      • Can I contribute to TorchSOM?
  • Troubleshooting
    • Installation Issues
      • Package Issues
      • CUDA/GPU Issues
      • CUDA not available
    • Training Problems
      • Training doesn’t converge
      • Very slow training
      • NaN values in results
    • Visualization Issues
      • Empty or white visualizations
      • Figures not displaying
      • Poor visualization quality
    • Data Issues
      • Poor clustering results
    • Configuration Errors
      • ValidationError from Pydantic
      • Parameter compatibility issues
    • Memory Issues
      • Memory usage too high
      • Memory leaks
    • Getting Help
      • Diagnostic Information
      • Creating Minimal Examples
      • Where to Get Help
      • Debug Mode
TorchSOM
  • Search

© Copyright 2025, Manufacture Française des Pneumatiques Michelin.

Built with Sphinx using a theme provided by Read the Docs.