0
$\begingroup$

LSTM is good for sequence prediction, because it can remember the previous context. What is the rationale behind using it in classification tasks ? In particular, they have used it for the following task:

The Large Movie Review Dataset (often referred to as the IMDB dataset) contains 25,000 highly-polar movie reviews (good or bad) for training and the same amount again for testing. The problem is to determine whether a given movie review has a positive or negative sentiment.

In other words, can the classification problem be reduced to sequence prediction problem in some way?

$\endgroup$
  • $\begingroup$ can you give an example for such a application? $\endgroup$ – Peter Jun 5 at 9:52
  • $\begingroup$ I updated my question with a link $\endgroup$ – Holmes.Sherlock Jun 5 at 10:12
  • $\begingroup$ Spam vs Ham classification problem, is also a good example for LSTM use case. Though it can be done using normal Logistic-Regression model, but best results can be expect using NLP. $\endgroup$ – vipin bansal Jun 5 at 10:27
  • $\begingroup$ What do you mean? $\endgroup$ – Holmes.Sherlock Jun 5 at 10:28
3
$\begingroup$

I'll go for you questions one by one:

What is the rationale behind using it in classification tasks?

LSTM Networks for classifcation tasks are mainly due to NLP. In particular, Sentiment Analysis. What is the sentiment/polarization of a tweet? By "understanding" natural language, a model can guess whether it is positive or negative towards a given issue. Since the sequence of words is fundamental in order to understand natural language, you need to capture the sequence of words in order to understand the text, that you then need to then estimate the sentiment.

--

Can the classification problem be reduced to sequence prediction problem in some way?

I think NLP made the two tasks very related to each other, even though in my opinion they can still be analytically separated, so I wouldn't say that one can be reduced to the other.

However let's say that, once you have implemented a model that is very good at estimating the sentiment of a text, you can then make it generate positive/negative texts. In other words, they are closely related (thanks to NLP) but they are not the same.

Hope this helps, otherwise let me know.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.