n_nfeat = ndatas.size(2)
ndatas = scaleEachUnitaryDatas(ndatas)
# trans-mean-sub
ndatas = centerDatas(ndatas)
print("size of the datas...", ndatas.size())
# switch to cuda
ndatas = ndatas.cuda()
labels = labels.cuda()
#MAP
lam = 10
model = GaussianModel(n_ways, lam)
model.initFromLabelledDatas()
alpha = 0.2
optim = MAP(alpha)
optim.verbose = False
optim.progressBar = True
acc_test = optim.loop(model, n_epochs=20)
print(
"final accuracy found {:0.2f} +- {:0.2f}".format(*(100 * x
for x in acc_test)))
ndatas = QRreduction(ndatas)
n_nfeat = ndatas.size(2)
ndatas = scaleEachUnitaryDatas(ndatas)
# trans-mean-sub
ndatas = centerDatas(ndatas)
print("size of the datas...", ndatas.size())
# switch to cuda
ndatas = ndatas.cuda()
labels = labels.cuda()
#MAP
lam = LAMBDA
model = GaussianModel(n_ways, lam)
model.initFromLabelledDatas()
alpha = ALPHA
optim = MAP(alpha)
optim.verbose=False
optim.progressBar=True
acc_test = optim.loop(model, n_epochs=N_STEPS)
print("final accuracy found {:0.2f} +- {:0.2f}".format(*(100*x for x in acc_test)))