mlpro 0.8.0

Machine Learning Professional

MLPro - Machine Learning Professional

Machine Learning Professional - A Synoptic Framework for Standardized Machine Learning Tasks in Python

MLPro was developed in 2021 by Automation Technology and Learning Systems team at Fachhochschule Südwestfalen

MLPro provides complete, standardized, and reusable functionalities to support your scientific research, educational task, or industrial project in machine learning.

In the first version of MLPro, we provide a standardized Python package for reinforcement learning (RL) and game theoretical (GT) approaches, including environments, algorithms, multi-agent RL (MARL), model-based RL (MBRL) and many more. Additionally, we incorporate the available third party packages by developing wrapper classes to enable our users to reuse the third party packages in MLPro.

Main Features

• Test-driven development (CI/CD concept)
• Clean code and constructed through Object-Oriented Programming
• Ready-to-use functionalities
• Usability in scientific, industrial and educational contexts
• Extensible, maintainable, understandable
• Attractive UI support (available soon)
• Reuse of available state-of-the-art implementations
• Clear documentations

Documentation

The Documentation is available on : https://mlpro.readthedocs.io/

Installation

Prerequisites

MLPro requires Python 3.7+

pip install mlpro

Example

This examples shows how to train CartPole-V1 with Stable-Baselines3 wrapper

import gym

from stable_baselines3 import PPO

from mlpro.rl.models import *
from mlpro.wrappers.openai_gym import WrEnvGYM2MLPro
from mlpro.wrappers.sb3 import WrPolicySB32MLPro

class MyScenario(RLScenario):

    C_NAME      = 'Matrix'

    def _setup(self, p_mode, p_ada, p_logging):
        gym_env     = gym.make('CartPole-v1')
        self._env   = WrEnvGYM2MLPro(gym_env, p_logging=p_logging) 

        policy_sb3 = PPO(
                    policy="MlpPolicy",
                    n_steps=5, 
                    env=None,
                    _init_setup_model=False)

        policy_wrapped = WrPolicySB32MLPro(
                p_sb3_policy=policy_sb3, 
                p_observation_space=self._env.get_state_space(),
                p_action_space=self._env.get_action_space(),
                p_ada=p_ada,
                p_logging=p_logging)
        
        return Agent(
            p_policy=policy_wrapped,   
            p_envmodel=None,
            p_name='Smith',
            p_ada=p_ada,
            p_logging=p_logging
        )

training = RLTraining(
        p_scenario_cls=MyScenario,
        p_cycle_limit=1000,
        p_max_adaptations=0,
        p_max_stagnations=0,
        p_visualize=True,
        p_logging=Log.C_LOG_ALL )

training.run()

Implemented Wrappers

Features	Status
OpenAI-Gym	:heavy_check_mark:
Stable-Baselines3	:heavy_check_mark:
PettingZoo	:heavy_check_mark:
Hyperopt	:heavy_check_mark:

Mainteners

MLPro is currently maintained by Detlef Arend, M Rizky Diprasetya, Steve Yuwono, William Budiatmadjaja

How to contribute

If you want to contribute, please read CONTRIBUTING.md