mimllearning 0.2.41

MIML Learning Library

miml: Multi Instance Multi Label Learning Library for Python

The aim of the library is to ease the development, testing and comparison of classification algorithms for multi-instance multi-label learning (MIML).

Table of Contents

• Installation
• Usage
• License

Installation

Use the package manager pip to install miml.

$ pip install mimllearning

Requirements

The requirement packages for miml library are: numpy, scikit-learn, scipy, mil, tensorflow and keras. Installing miml with the package manager does not install the package dependencies. So install them with the package manager manually if not already downloaded.

$ pip install numpy
$ pip install scikit-learn
$ pip install scipy
$ pip install mil
$ pip install tensorflow
$ pip install keras==2.12.0

Usage

Datasets

import pkg_resources
from miml.data.load_datasets import load_dataset

dataset_train = load_dataset(pkg_resources.resource_filename('miml', 'datasets/miml_birds_random_80train.arff'),
                             delimiter="'")
dataset_test = load_dataset(pkg_resources.resource_filename('miml', 'datasets/miml_birds_random_20test.arff'),
                            delimiter="'")

Classifier

from miml.classifier import MIMLtoMIBRClassifier, AllPositiveAPRClassifier

classifier_mi = MIMLtoMIBRClassifier(AllPositiveAPRClassifier())
classifier_mi.fit(dataset_train)
results_mi=classifier_mi.evaluate(dataset_test)

Report

from miml.report import Report

report = Report()
report.to_string(dataset_test.get_labels_by_bag(), results_ml)
print("")
report.to_csv(dataset_test.get_labels_by_bag(), results_ml)

License

• MIT license