CODEWITHSUNDEEP
machine-learningneural-network
Ishpreet
Ishpreet

Reputation: 5880

Can a neural network have integer inputs?

I build a neural network with input as a mixture of integers and booleans. But it did not converge. I have seen many examples on internet and every one of them has input in boolean form. So is it possible to build a neural network with a mixture of inputs or integer inputs?

Upvotes: 2

Views: 5608

Answers (1)

Sahil
Sahil

Reputation: 1386

Indeed, it is. What you probability need to do is to normalize your inputs. This means that you could divide a feature's value by the maximum value you expect to see in that place, so that everything lies in the range (-1,1).

Some links to understand normalization of inputs: Why do we have to normalize the input for an artificial neural network? https://www.researchgate.net/post/How_can_I_normalize_input_and_output_data_in_training_neural_networks

Another recent way to ensure normalization is the concept of batch normalization

Upvotes: 2

Related Questions