sevq 1.0.1

SEVQ: Simplified Evolving Vector Quantization

sevq

SEVQ: Simplified Evolving Vector Quantization

• Free software: MIT license

• Documentation: https://sevq.readthedocs.io.

Installation

To install sevq, run this command in your terminal:

pip install sevq

Usage

Training and prediction one sample at a time

from sevq.algorithm import SEVQ

c = SEVQ()
c.partial_fit([-2, -2], 2)
c.partial_fit([-1, -1], 1)
c.partial_fit([1, 1], 1)
c.partial_fit([2, 2], 2)

print(c.predict([0, 0]))  # 1
print(c.predict([3, 3]))  # 2
print(c.predict([-3, -3]))  # 2

Training and prediction on multiple samples

from sevq.algorithm import SEVQ

c = SEVQ()
c.fit(
    [[-2, -2], [-1, -1], [1, 1], [2, 2]],
    [2, 1, 1, 2],
    epochs=1, permute=False
)

print(c.predict([[0, 0], [3, 3], [-3, -3]]))  # [1, 2, 2]

To replicate the research run:

# python 3.6 is required by neupy
git clone https://github.com/sylwekczmil/sevq
cd sevq
# create venv and activate
# up to you

# install dependencies
pip3 install -r requirements_research.txt

# run research code, please ignore warnings, this script can run long time
python3 run_research.py

History

1.0.0 (2021-11-24)

• First release on PyPI.

1.0.1 (2021-11-24)

• Fix typos in release.