Review_Classification_IMDB

Assignment for sentiment classification of the reviews from IMDB dataset: https://ai.stanford.edu/~amaas/data/sentiment/

Classifier

The classifier SentimentClassifier() consists of:

self.emblayer = nn.Embedding(self.config.VOC_SIZE, self.config.EMBED_SIZE)
self.lstmlayer = nn.LSTM(self.config.EMBED_SIZE, self.config.HIDDEN_SIZE ,batch_first=True)
self.linear1 = nn.Linear(self.config.HIDDEN_SIZE, 32)
self.relu = nn.ReLU()
self.linear2 = nn.Linear(32, 2)

Embedding layer

The nn.Embedding layer in PyTorch is a layer that maps discrete categorical variables to continuous vector representations, known as "embeddings". It is commonly used in natural language processing and other text-based tasks to represent words as dense, continuous vectors, which are easier to work with than the one-hot encoded representations that are commonly used as inputs to traditional neural networks.

This model takes in a batch of sequences of word indices of shape (batch_size, seq_length) and passes them through the embedding layer to produce a batch of sequences of word embeddings of shape (batch_size, seq_length, embedding_dim).

LSTM layer

Long Short-Term Memory (LSTM) is a type of recurrent neural network (RNN) that is able to capture long-term dependencies in sequential data. It is called "long short-term memory" because it can remember information for long periods of time, but it is also able to forget irrelevant information and keep the most important information.

An LSTM layer nn.LSTM in PyTorch consists of four different "gates", which control the flow of information into and out of the memory cell:

The "input gate" determines which parts of the input to store in the memory cell.
The "forget gate" determines which parts of the previous state to forget.
The "output gate" determines which parts of the previous state to output.
The "memory cell" stores the relevant parts of the input and the previous state.

At each time step, the LSTM layer takes in an input x and the previous hidden state h and updates the hidden state based on these gates and the memory cell. The updated hidden state is then output and used as the input to the next time step.

How to run this code

You'll need Git to be installed on your computer.

# Clone this repository
$ git clone https://github.com/natasabrisudova/Review_Classification

To try the model run in command prompt:

$ predict.py ReviewToBeClassified

For example:

$ predict.py I love this movie

# Review: I love this movie 
# Classification: Positive