README.md

multiml

multiml is a prototype framework for developing multi-step machine learnings. The framework provides a method to connect multiple machine learnings with features of:

• optimizing combination of machine learning models and hyperparameters,
• extensibility that users can implement new algorithms.

Any new idea, issues and Pull requests are welcome!

Installation

Requiremnets:

• CentOS 7.6+
• Python 3.8+

minimum setup

$ pip install -e .

with machine learning libraries

$ pip install -e .[pytorch,tensorflow]

Quick start and API references

Please see multiml documentation.

Tests and coding style

Please test and apply yapf before you commit changes.

$ python setup.py test

$ yapf -i [changed file]

Overview of multiai framework

Examples

See the example directory. The following example is to run two-step machine learnings using iris dataset.

$ python run_iris_classification.py

Application

• Event Classification with Multi-step Machine Learning(arxiv)