Neuraxio
/
Neuraxle
0.5.4
Contents:
Introduction
Encapsulate Models and Data Transformers
Pipe and Filter
Features
Wrapper (a.k.a. Decorator) classes
Pipelines for Minibatching and Parallel Processing
Repository for lazy data loading
Training your pipeline
Serializing your pipeline
Conclusion
Neuraxle Pipelines
Documentation
Installation
Examples
Why Neuraxle ?
Community
License
Citation
Contributors
Supported By
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
Handler Methods
handle_fit_transform
handle_fit
handle_transform
When to use handler methods ?
HandleOnlyMixin
ForceHandleMixin
ForceHandleOnlyMixin
Examples
ForEachDataInput
ToNumpy
Transform Expected Outputs
Expand The DataContainer
Reversible Pipeline
Introduction to Automatic Hyperparameter Tuning
AutoML loop
1. Define your pipeline
2. Choose a validation splitter
3. Define a the main scoring metric with ScoringCallback
4. Add metric callbacks with MetricCallback (optional)
5. Select an hyperparams repository
6. Select an hyperparams optimizer
7. Create, and launch AutoML loop
Run 10 trials
Get best model, and predict
Introduction to Random Distributions
Plotting Each Hyperparameter Distribution
Discrete Distributions
RandInt
Boolean
Choice
Priority Choice
Continuous Distributions
Continuous Uniform
Continuous Loguniform
Continuous Normal
Continuous Lognormal
Continuous Normal Clipped
Continuous Lognormal Clipped
Quantized Hyperparameter Distributions
Quantized Uniform
Repaired Quantized Uniform
Quantized Log Uniform
Quantized Normal
Quantized Lognormal
Creating your own distributions
Using Scipy Distributions
Creating your own distributions using scipy
BaseCustomContinuousScipyDistribution
BaseCustomDiscreteScipyDistribution
Step Saving And Lifecycle
Lifecycle
Step Saving
Saver
Custom Saver Example
Saving Example
Pipeline
Full Dump Saving
Full Dump Loading
REST API Serving
Import Packages
Load your Dataset
Create your Pipeline
Let’s Train and Test
Deploy the Pipeline
Write a step to decode the accepted JSON as data inputs
Write a step to encode the returned JSON response
Finally Serve Predictions
API Call Example
Next Steps
Pipeline Serialization
Data Transformation Caching
Checkpoints
Complete API Documentation
neuraxle.base
Neuraxle’s Base Classes
neuraxle.pipeline
Neuraxle’s Pipeline Classes
neuraxle.data_container
Neuraxle’s DataContainer classes
neuraxle.union
Union of Features
neuraxle.checkpoints
Neuraxle’s Checkpoint Classes
neuraxle.api
Neuraxle’s API classes
neuraxle.metrics
Neuraxle’s metrics classes
neuraxle.steps.numpy
Pipeline Steps Based on NumPy
neuraxle.steps.flow
Neuraxle’s Flow Steps
neuraxle.steps.data
Data Steps
neuraxle.steps.column_transformer
Neuraxle’s Column Transformer Steps
neuraxle.steps.features
Featurization Steps
neuraxle.steps.sklearn
Pipeline Steps Based on Scikit-Learn
neuraxle.steps.loop
Pipeline Steps For Looping
neuraxle.steps.caching
Pipeline Steps For Caching
neuraxle.steps.output_handlers
Output Handlers Steps
neuraxle.steps.misc
Miscelaneous Pipeline Steps
neuraxle.hyperparams.distributions
Hyperparameter Distributions
neuraxle.hyperparams.space
Hyperparameter Dictionary Conversions
neuraxle.metaopt.auto_ml
Neuraxle’s AutoML Classes
neuraxle.metaopt.trial
Neuraxle’s Trial Classes
neuraxle.metaopt.callbacks
Neuraxle’s training callbacks classes.
neuraxle.metaopt.random
Random
neuraxle.metaopt.sklearn
Scikit-learn metaoptimizers
neuraxle.metaopt.tpe
Tree parzen estimator
neuraxle.metaopt.deprecated
Neuraxle’s Automatic Machine Learning Classes
neuraxle.distributed.streaming
Streaming Parallel Data Processing
neuraxle.rest.flask
Neuraxle’s Flask Wrapper classes
Usage Examples
AutoML
Usage of AutoML loop, and hyperparams with sklearn models.
Caching
Usage of ValueCachingWrapper in Neuraxle.
Usage of Checkpoints in Automatic Machine Learning (AutoML)
REST API Model Serving
Easy REST API Model Serving with Neuraxle
Getting started
Inverse Transforms in Neuraxle: How to Reverse a Prediction
Create Nested Pipelines in Neuraxle
Create Pipeline Steps in Neuraxle that doesn’t fit or transform
Create label encoder across multiple columns
Create Pipeline Steps that require implementing only handler methods
Hyperparameters
Manipulate Hyperparameter Spaces for Hyperparameter Tuning
Parallel
Parallel processing in Neuraxle
Neuraxle hyperparameter examples
Boston Housing Regression
Boston Housing Regression with Meta Optimization
Neuraxio
/
Neuraxle
Docs
»
Search
Please activate JavaScript to enable the search functionality.
