I have the same preference as you do and the pep8 validator doesn't think it's wrong: https://pypi.python.org/pypi/pep8
Usually I think it's recommended to do it like this however, I find it less readable:
emissionprob = preprocessing.normalize(self.random_state.rand(self.n_components,
self.n_symbols),
norm='l1',
axis=1,
copy=False)
I personally do it like this (note the trailing comma's everywhere):
emissionprob = preprocessing.normalize(
self.random_state.rand(
self.n_components,
self.n_symbols,
),
norm='l1',
axis=1,
copy=False,
)