embayes 0.1.1

Naive Bayes classifier for embedded systems

# embayes Bayesian machine learning classifiers for embedded systems. Train in Python, deploy on microcontroller.

## Key features

Embedded-friendly Classifier

• Portable C99 code

• No stdlib required

• No dynamic allocations

• Integer/fixed-point math only

• Single header file include

• Fast, sub-millisecond classification

Convenient Training

• API-compatible with [scikit-learn](http://scikit-learn.org)

• Implemented in Python 3

• C classifier accessible in Python using pybind11

[MIT licensed](./LICENSE.md)

## Status Minimally useful

• Gaussian Naive Bayes classifier implemented

• Tested running on ESP8266 and Linux.

• On ESP8266, 2 classes and 30 features classify in under 0.5ms

## Installing

Install from git

git clone https://github.com/jonnor/embayes python3 setup.py install –user

## Usage

See [examples/cancer.py](./examples/cancer.py) and [embayes.ino](./embayes.ino)

## TODO

0.2

• Make estimator a wrapper around sklearn.naivebayes.GaussianNB

• Make estimator work in sklearn pipeline

• Make pdf approximation configurable as parameter

1.0

• Support generating inline C code, not needing model coefficients in RAM

• Support de/serializing coefficients at runtime

• Support training on microcontroller