Skip to content

Latest commit

 

History

History
48 lines (36 loc) · 887 Bytes

README.md

File metadata and controls

48 lines (36 loc) · 887 Bytes

Ziptie: An unsupervised feature learning algorithm

Here's the full story of how Ziptie works and why it was created.

Installation

First download the code. In a terminal with a bash-like scripting language this can be done with

git clone https://codeberg.org/brohrer/ziptie.git

Then Python's package manager pip is a good way to get it into your environment.

python3 -m pip install -e ziptie

Usage

Import

from ziptie.algo import Ziptie

Initialize

zt = Ziptie(n_inputs)

Update clusters on each iteration

zt.create_new_bundles()
zt.grow_bundles()

Calculate outputs on each iteration

bundle_activities = zt.update_bundles(inputs)

Impose a stopping condition

zt.n_bundles >= bundle_limit