def run_single(run_id):
'''
Given an input dictionary containing a single paramater set
run mcmc in matlab using bs_macros.run_matlab.
'''
input_dict = butils.load_data(run_id,'input')
return bsm.runmat('run_mcmc', input_dict, run_id)
def compute_embedding(affinities,
aff_type = 'pairs',
do_mve = False,
ss_multiplier = None):
mve_dict = dict(k = 4,
similarities = affinities,
do_mve = False if aff_type == 'easy' else False)
embedding= bsm.runmat('mve_frompy', mve_dict, 'subopt_mve')
mve_vecs = embedding['Y'].T
pca_vecs = embedding['YPCA'].T
return pca_vecs, mve_vecs
def test_bsubfun(run_id):
"""
A sample function to demonstrate the calling of a matlab script (here,
ap_frompy) from within python. Taking an input dictionary and a run_id,
this script is designed to be called using the 'eyeball' class from
utils/bsub.py.
inputs:
input_dict: {similarities: a similarity matrix for the input points,
self_similarity: a single value for the self similarity
of datapoints. Control cluster size.
outputs:
outpt_dict: {indexes: cluster exemplar indices.}
"""
input_dict = butils.load_data(run_id, "input")
return bsm.runmat("ap_frompy", input_dict, run_id)
def test_bsubfun(run_id):
'''
A sample function to demonstrate the calling of a matlab script (here,
ap_frompy) from within python. Taking an input dictionary and a run_id,
this script is designed to be called using the 'eyeball' class from
utils/bsub.py.
inputs:
input_dict: {similarities: a similarity matrix for the input points,
self_similarity: a single value for the self similarity
of datapoints. Control cluster size.
outputs:
outpt_dict: {indexes: cluster exemplar indices.}
'''
input_dict = butils.load_data(run_id, 'input')
return bsm.runmat('ap_frompy', input_dict, run_id)
def bic_clustering(run_id):
"""
A matlab/bsub process to compute the BIC maximal clustering for an
input dictionary containing a similarity matrix.
inputs:
input_dict: {similarities: a similarity matrix}
outputs:
output_dict: {inds:cluster exemplar indices, (MAX BIC)
self_similarity:float, self similarity (MAX BIC)
inds_[#]: (same as above, ALL BIC)
self_similarity_[#}: (...)
bic_[#]: (...)
}
"""
input_dict = butils.load_data(run_id, "input")
return bsm.runmat("ap_max_bic", input_dict, run_id)
def bic_clustering(run_id):
'''
A matlab/bsub process to compute the BIC maximal clustering for an
input dictionary containing a similarity matrix.
inputs:
input_dict: {similarities: a similarity matrix}
outputs:
output_dict: {inds:cluster exemplar indices, (MAX BIC)
self_similarity:float, self similarity (MAX BIC)
inds_[#]: (same as above, ALL BIC)
self_similarity_[#}: (...)
bic_[#]: (...)
}
'''
input_dict = butils.load_data(run_id, 'input')
return bsm.runmat('ap_max_bic', input_dict, run_id)
def compute_clusters(affinities, ss):
cluster_dict = dict(similarities=affinities,
self_similarity = ss)
clusters = bsm.runmat('ap_frompy', cluster_dict, 'subopt_clusters')
return squeeze(clusters['inds'])