IMDB-experiments

Text Classification experiments on IMDB dataset using pretrained word embeddings (Word2Vec, Glove, FasText ) and Neural Network architecture like like CNN, RNN.

• Refer https://saurabhk30.medium.com/text-classification-on-imdb-dataset-using-word-embedding-5ce301b23816 blog post form details.

Steps followed for Text Classification

Reported Model Performances in the Experiment

• Movie data preprocessing details about text preprocessing steps.

• The value mentioned in Sr.No columnn the () corresponsds to the experiment in IMDB-experiments/Expt with embeddings.

• 2027 Out of vocabulary word found in Fastext Embeding

• 327 Out of vocabulary word found in Glove

• When embeddings layers parameter trainable parameter is False model has poor generalisation.