About Neuraxle

• Neuraxle Pipelines
  • Documentation
  • Installation
    • Examples
      • Why Neuraxle ?
      • Community
    • License
    • Citation
    • Contributors
    • Supported By
• Comparison to Other Machine Learning Pipeline Frameworks, and Compatibility
  • scikit-learn
  • Apache Beam
  • spaCy
  • Kubeflow
  • TensorFlow
  • Hyperopt
• Solutions to Scikit-Learn’s Biggest Problems
  • Definitions
  • Inability to Reasonably do Automatic Machine Learning (AutoML)
    • Problem: Defining the Search Space (Hyperparameter Distributions)
      • Solution: Define Hyperparameter Spaces Within the Steps
    • Problem: Defining Hyperparameters in the Constructor is Limiting
      • Solution: Separate Steps’s Constructors From the get_params Method
    • Problem: Different Train and Test Behavior
      • Solution: use the Set Train Special Method and use Step Wrappers
    • Problem: You trained a Pipeline and You Want Feedback Statistics on its Learning
      • Solution: the Introspect Special Method
  • Inability to Reasonably do Deep Learning Pipelines
    • Problem: Scikit-Learn Hardly Allows for Mini-Batch Gradient Descent (Incremental Fit)
      • Solution: Minibatch Pipeline Class and the Ability to Incrementally Fit Pipeline Steps
    • Problem: Initializing the Pipeline and Deallocating Resources
      • Solution: Add Setup and Teardown Lifecycle Methods to Your Steps
    • Problem: It is Difficult to Use Other Deep Learning (DL) Libraries in Scikit-Learn
      • Solution: Moar Steps Lifecycle Methods
    • Problem: The Ability to Transform Output Labels
      • Solution: OutputTransformerWrapper and InputAndOutputTransformerMixin
  • Not ready for Production nor for Complex Pipelines
    • Problem: Processing 3D, 4D, or ND Data in your Pipeline with Steps Made for Lower-Dimensionnal Data
      • Solution: use a ForEachDataInputs Wrapper to Loop from ND Data to N(D-1) Data
    • Problem: Modify a Pipeline Along the Way, such as for Pre-Training or Fine-Tuning
      • Solution: the Mutate Special Method
      • Another Solution: the Apply Special Method
    • Problem: Getting Model Attributes from Scikit-Learn Pipeline
      • Solution: Simpler Nested Pipelines __getitem__ Methods
    • Problem: You can’t Parallelize nor Save Pipelines Using Steps that Can’t be Serialized “as-is” by Joblib
      • Solution: Use a Chain of Savers in each Step
      • About Cluster Computing and Parallelism in Python
• Awesome Neuraxle
  • Contents
  • Examples & Articles
  • Courses & Training
  • Videos
  • Projects
  • Community
  • License
• Contributing to Neuraxle
  • First steps
  • Before coding
  • Pull Requests
  • Code Reviews
  • Reviewing other’s code
  • Publishing project to PyPI
• Code of conduct
  • Respect
  • Politeness
  • Code reviews
• License
  • Summary of the License