pyism 0.0.2

A scikit-multiflow API for Tensorflow-based models implementing Incremental Sequence Models (ISM)

pyISM - Incremental Sequence Models for Python

What is pyISM?

The pyISM is a Python package implementing a family of Incremental Sequence Models (ISMs) implemented with tensorflow with an scikit-multiflow API.

How to install?

PyPI:

pip install -U pyism

GitHub:

pip install -U git+https://github.com/alvarolemos/pyism

What are Incremental Sequence Models (ISMs)?

Incremental Sequence Models are a family of Recurrent Neural Networks (RNNs) to tackle evolving data streams, aiming at less complexity without losing performance. It was proposed in the research An Incremental Learning approach using Sequence Models, to be presented by Álvaro Conde Lemos Neto as his Master's degree final dissertation in the Graduate Program in Electrical Engineering (PPGEE) of the Federal Univeristy of Minas Gerais (UFMG), supervisde by Prof. Criatiano Leite de Castro, PhD, both from the Machine Intelligence And Data Science (MINDS) lab.