/micromlgen

Generate C code for microcontrollers from Python's sklearn classifiers

Primary LanguagePythonMIT LicenseMIT

Introducing MicroML

MicroML is an attempt to bring Machine Learning algorithms to microcontrollers. Please refer to this blog post to an introduction to the topic.

Install

pip install micromlgen

Supported classifiers

micromlgen can port to plain C many types of classifiers:

  • DecisionTree
  • RandomForest
  • XGBoost
  • GaussianNB
  • Support Vector Machines (SVC and OneClassSVM)
  • Relevant Vector Machines (from skbayes.rvm_ard_models package)
  • SEFR
  • PCA
from micromlgen import port
from sklearn.svm import SVC
from sklearn.datasets import load_iris


if __name__ == '__main__':
    iris = load_iris()
    X = iris.data
    y = iris.target
    clf = SVC(kernel='linear').fit(X, y)
    print(port(clf))

You may pass a classmap to get readable class names in the ported code

from micromlgen import port
from sklearn.svm import SVC
from sklearn.datasets import load_iris


if __name__ == '__main__':
    iris = load_iris()
    X = iris.data
    y = iris.target
    clf = SVC(kernel='linear').fit(X, y)
    print(port(clf, classmap={
        0: 'setosa',
        1: 'virginica',
        2: 'versicolor'
    }))

PCA

It can export a PCA transformer.

from sklearn.decomposition import PCA
from sklearn.datasets import load_iris
from micromlgen import port

if __name__ == '__main__':
    X = load_iris().data
    pca = PCA(n_components=2, whiten=False).fit(X)
    
    print(port(pca))

SEFR

Read the post about SEFR.

pip install sefr
from sefr import SEFR
from micromlgen import port


clf = SEFR()
clf.fit(X, y)
print(port(clf))

DecisionTreeRegressor and RandomForestRegressor

pip install micromlgen>=1.1.26
from sklearn.datasets import load_boston
from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import RandomForestRegressor
from micromlgen import port


if __name__ == '__main__':
    X, y = load_boston(return_X_y=True)
    regr = DecisionTreeRegressor(max_depth=10, min_samples_leaf=5).fit(X, y)
    regr = RandomForestRegressor(n_estimators=10, max_depth=10, min_samples_leaf=5).fit(X, y)
    
    with open('RandomForestRegressor.h', 'w') as file:
        file.write(port(regr))
// Arduino sketch
#include "RandomForestRegressor.h"

Eloquent::ML::Port::RandomForestRegressor regressor;
float X[] = {...};


void setup() {
}

void loop() {
    float y_pred = regressor.predict(X);
}