CODEWITHSUNDEEP
pythonmachine-learningkeraslstm
user291678
user291678

Reputation: 111

Is it possible for LSTM to output a list of probabilities from a given list of possible outputs?

as the title says really. I've managed to implement a very simple LSTM model that takes an input sequence and outputs a given float value.

Firstly, I'd like to know if it's possible to get answers taken from a set of possible answers. E.g. if I know the answer should be in [1,2,3] to output the answer as being 1.

Secondly, if possible I'd like the output to be a probability distribution on the possible answers, e.g. [0.5,0.3,0.2].

I've implemented my simple LSTM model in Python using the various keras packages. Any pointers to the right direction to learn about how to implement this would be great!

Upvotes: 1

Views: 837

Answers (1)

Sahil Desai
Sahil Desai

Reputation: 3696

LSTM is basically one type of recurrent neural network which provide many to many functionality. For that you need to add final dense layer with same number of input class in softmax layer, so you will get exact probability for each input class.

Upvotes: 3

Related Questions