Hopfield neural network for clustering

Hello,

I would like to use the Hopfield network for clustering of a data sample.

The sample consists of vectors 1x15 and are all normalized in the range [0,1].

The built in "newhop" function requires target vectors with binary values (-1 or 1).

-Can I use random initial weights for the neurons?

-Can I define target vectors with no binary values, i.e. selected from the data sample?

Thank you very much.

Pavlos