Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python.announce > #2607 > unrolled thread

Conx, Easy to use Deep Learning

Back to article view | Back to comp.lang.python.announce

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Conx, Easy to use Deep Learning Doug Blank <doug.blank@gmail.com> - 2017-08-19 08:56 -0400

#2607 — Conx, Easy to use Deep Learning

An initial release of new Python library for Deep Learning: conx (pronounced
"connects"). Built on Keras, but with enhancements, and designed for
researchers, teaching, and learning.

Documentation: http://conx.readthedocs.io/
Installation: https://github.com/Calysto/conx

What niche does conx fill?

* designed for students and researchers that don't want to work with (or
don't know) numpy, matplotlib, or Keras functional-api
* tight integration with Jupyter notebooks
* widget-based app for easy exploration of ANN models
* error messages designed for humans
* standard input and target dataset API (separate from internal Keras
numpy-based format)
* supports all Keras layers via thin wrappers
* creates standard Keras models
* extended propagate functions: propagate_to, propagate_from
* designed for use with an online robot simulator (optional)
* built-in plotting and webcam interfaces
* creates updatable network images (SVG) with activations, like this one:

https://raw.githubusercontent.com/Calysto/conx/master/notebooks/network.png

We're just getting started, so if you would like to collaborate, or have
ideas, please let us know!

Discussion group here: https://groups.google.com/forum/#!forum/conx-users

[toc] | [standalone]

Back to top | Article view | comp.lang.python.announce

csiph-web