Folders and files Name Name Last commit message
Last commit date
parent directory
View all files
Sentiment Analysis using RNN:
Sentiment analysis is typically a classification problem.
However for a larger data set RNN can provide a better prediction than typical Machine Learning approach.
This is training vs validation loss.
Since we have around 25K reviews, the network is over-fitting beyond 5 epochs
We need to calculate the loss against only the last sigmoid unit.
The embedding layer will be trained automatically.
Input Dataset : 25K Reviews
Train/Validation/Test Split : 80%/10%/10%
Number of hidden layers : 256
LSTM Layer ( stacking ) : 2
Learning Rate : 0.001
Batch Length : 50
epochs : 4
Average Train Loss : 0.6725
Average Validation Loss : 1.2836
Average Test Loss : 0.529
Extract the reviews.zip in the dataset folder
Here is another work for Sentiment Analysis using Machine Leaning.
Visualization:
Machine Leaning Model:
TBA (To be added to github)
You can’t perform that action at this time.
