I wanted to apply LSTM on a numerical dataset for binary classification problems. First of all, I apply the word embedding method to convert text into the numerical dataset. Then the dataset is like the below table:
Now how do I arrange my dataset for applying LSTM since LSTM is applied on sequential data?
Can anyone please help me?
Do you want to implement that algorithm from scratch or are you fine with using frameworks? Here's for instance a pretty straight forward tutorial for binary sentiment analysis that uses sklearn and Keras to train an LSTM on GloVe embeddings:
https://towardsdatascience.com/sentiment-analysis-using-lstm-and-glove-embeddings-99223a87fe8e
You can use the last output of LSTM and then feed into linear layer with input features the same as hidden size and output equals to the number of class (in your case: 2) and then apply the sigmoid activation function.
You can follow this blog https://www.analyticsvidhya.com/blog/2021/06/lstm-for-text-classification/
this link may help
https://stackoverflow.com/questions/60987125/can-we-apply-lstm-rnn-for-binay-classification-on-numeric-values-data
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning