view README.txt @ 66:9f1324a59cc1

add bounding box + tests
author Jeff Hammel <k0scist@gmail.com>
date Sun, 17 Dec 2017 12:48:06 -0800
parents 2adeb95cf4d5
children
line wrap: on
line source

tvii
====

TV2 :  experiments in machine learning


Dependencies
------------

`tvii` is built upon the shoulders of giants

- numpy: http://www.numpy.org/
- h5py: http://www.h5py.org/
- matplotlib: http://matplotlib.org/
- scipy: https://www.scipy.org/install.html
- sci-kit learn: http://scikit-learn.org/stable/install.html

For `h5py` you may need your `python-dev` package
and or `libhdf5-dev`. For `scipy` you need Lapack.

You may install these with python 3.6 using::

   pip install -e .


Useful links + software
-----------------------

- [tensorflow](https://www.tensorflow.org/)
  * [Parsey McParseface](https://github.com/tensorflow/models/tree/master/syntaxnet#using-the-pre-trained-nlp-models) (SyntaxNet)
- [torch](http://torch.ch/)
- [spaCy](https://spacy.io/) (NLP)
- [NakedTensor] (https://github.com/jostmey/NakedTensor) :  Bare bone examples of machine learning in TensorFlow
- [Keras](https://keras.io/) : Keras is a high-level neural networks API, written in Python and capable of running on top of TensorFlow, CNTK, or Theano
- [Caffe](http://caffe.berkeleyvision.org/)
- [PaddlePaddle](http://doc.paddlepaddle.org/release_doc/0.9.0/doc/introduction/index.html): PaddlePaddle is a deep learning platform open-sourced by Baidu


Ideas
-----

* Given: a corpus of text, for a given acronym, yield
what it stands for (Bonus points:  likely error rate and other
high-value matches)


Links
-----

Things to check out:

- mxnet : http://gluon.mxnet.io/P01-C01-introduction.html

- NakedTensor: https://github.com/jostmey/NakedTensor;
  Bare bone examples of machine learning in TensorFlow

- http://pybrain.org/

- http://deeplearning.net/software/theano/

- https://spacy.io : Industrial-Strength Natural Language Processing

----

Jeff Hammel

http://k0s.org/hg/tvii