movie_review, a keras code which sets up a neural network to classify movie reviews from the IMDB database. The code seeks a correlation between the occurrence of certain keywords and the movie review being reported as positive or negative. Once a relationship is inferred between keywords and types of review, new reviews are classified using this same criterion.
This script references the keras imdb dataset. By default, this is stored in the location ~/.keras/datasets, and involves two files, imdb.npz, and imdb_word_index.json. If these files are not there at run time, keras will attempt to download them to that location from a server, before proceeding.
The script originally referenced the 10,000 most commonly used words. When running on my laptop, I was running out of memory. Reducing to the 5,000 most commonly used words allowed my computation to complete.
The information on this web page is distributed under the MIT license.
imdb, a keras code which sets up a neural network to classify movie reviews as positive or negative. This is similar to the "movie_review" script, but is designed to be run on a remote computing node which does not have internet access at run time.
imdb_dataset_load, a keras code which can download a copy of the IMDB dataset and place it in the default location if that has not already been done.
imdb_decode, a keras code which shows how to decode an IMDB dataset movie review.
newswire, a keras code which sets up a neural network for the classification of Reuters newswire texts into one of 46 categories.
Original keras version by Francois Chollet; This version by John Burkardt.
