/text-classification

Classifying text with Tensorflow and Keras

Primary LanguagePython

Text Classification With Keras

This repository contains code showing different methods of classifying text with the use of Tensorflow and Keras.

Data

I've taken the data from this dataset and consolidated the text and labels in a single TSV file. Note that the text needs to be cleaned of all special characters (@#!$% etc). After that's done I placed the file in the data folder with the name out.tsv.

Dense

This method uses a series of simple Dense and PReLU layers with dropouts in between. Specifically, it starts with

out = Dense(512, activation='relu', input_shape=(max_words,))(inputs)
out = PReLU()(out)
out = Dropout(0.25)(out)

Then I've added three sets of hidden layers with decreasing units and dropout amount and ends with a dense layer which returns the expected output shape.