mgplvm.rdist.relie module

class mgplvm.rdist.relie.ReLie(manif, m, n_samples, kmax=5, sigma=1.5, gamma=None, fixed_gamma=False, diagonal=False, mu=None, initialization='random', Y=None)[source]

Bases: mgplvm.rdist.relie.ReLieBase

name = 'ReLie'

property prms

training: bool

class mgplvm.rdist.relie.ReLieBase(manif, f, kmax=5, diagonal=False)[source]

Bases: mgplvm.rdist.common.Rdist

concentration_parameters()[source]

gmu_parameters()[source]

lat_gamma(Y=None, batch_idxs=None, sample_idxs=None)[source]

lat_gmu(Y=None, batch_idxs=None, sample_idxs=None)[source]

lat_prms(Y=None, batch_idxs=None, sample_idxs=None)[source]

msg(Y=None, batch_idxs=None, sample_idxs=None)[source]

mvn(gamma=None)[source]

name = 'ReLieBase'

property prms

sample(size, Y=None, batch_idxs=None, sample_idxs=None, kmax=5, analytic_kl=False, prior=None)[source]: generate samples and computes its log entropy

training: bool