def job_scf_gwgrid(sample_source, tr, te, r):
with util.ContextTimer() as t:
rand_state = np.random.get_state()
np.random.seed(r + 92856)
d = tr.dim()
T_randn = np.random.randn(J, d)
np.random.set_state(rand_state)
# grid search to determine the initial gwidth
mean_sd = tr.mean_std()
scales = 2.0**np.linspace(-4, 4, 20)
list_gwidth = np.hstack(
(mean_sd * scales * (d**0.5), 2**np.linspace(-8, 8, 20)))
list_gwidth.sort()
besti, powers = tst.SmoothCFTest.grid_search_gwidth(
tr, T_randn, list_gwidth, alpha)
# initialize with the best width from the grid search
best_width = list_gwidth[besti]
scf_gwgrid = tst.SmoothCFTest(T_randn, best_width, alpha)
test_result = scf_gwgrid.perform_test(te)
result = {
'test_method': scf_gwgrid,
'test_result': test_result,
'time_secs': t.secs
}
return result
def job_scf_gwopt(prob_label, tr, te, r, ni, n):
"""SmoothCFTest. Optimize only the Gaussian width.
Fix the test frequencies"""
raise ValueError('Use job_scf_gwgrid instead')
with util.ContextTimer() as t:
op_gwidth = {
'max_iter': 200,
'gwidth_step_size': 0.1,
'batch_proportion': 1.0,
'tol_fun': 1e-3
}
# optimize on the training set
rand_state = np.random.get_state()
np.random.seed(seed=r + 92856)
ss, _ = get_sample_source(prob_label)
T_randn = np.random.randn(J, ss.dim())
np.random.set_state(rand_state)
gwidth, info = tst.SmoothCFTest.optimize_gwidth(
tr, T_randn, **op_gwidth)
scf_gwopt = tst.SmoothCFTest(T_randn, gwidth, alpha)
return {
#'test_method': scf_gwopt,
'test_result': scf_gwopt.perform_test(te),
'time_secs': t.secs
}
def job_scf_opt10(sample_source, tr, te, r):
"""SmoothCFTest with frequencies optimized."""
op = {'n_test_freqs': J, 'max_iter': 100, 'freqs_step_size': 1.0,
'gwidth_step_size': 0.1, 'seed': r+92856, 'tol_fun': 1e-3}
test_freqs, gwidth, info = tst.SmoothCFTest.optimize_freqs_width(tr, alpha, **op)
scf_opt = tst.SmoothCFTest(test_freqs, gwidth, alpha)
scf_opt_test = scf_opt.perform_test(te)
return scf_opt_test
def job_scf_opt(sample_source, tr, te, r):
"""SmoothCFTest with frequencies optimized."""
with util.ContextTimer() as t:
op = {'n_test_freqs': J, 'max_iter': 500, 'freqs_step_size': 1.0,
'gwidth_step_size': 0.1, 'seed': r+92856, 'tol_fun': 1e-3}
test_freqs, gwidth, info = tst.SmoothCFTest.optimize_freqs_width(tr, alpha, **op)
scf_opt = tst.SmoothCFTest(test_freqs, gwidth, alpha)
scf_opt_test = scf_opt.perform_test(te)
result = {'test_method': scf_opt, 'test_result': scf_opt_test, 'time_secs': t.secs}
return result
def test(self, X, Y):
XY = self.preprocess(X, Y)
train, test = XY.split_tr_te(tr_proportion=self.split_ratio)
with contextlib.redirect_stdout(None):
with contextlib.redirect_stderr(None):
test_freqs, gwidth, info = fot_tst.SmoothCFTest.optimize_freqs_width(
train, self.alpha, n_test_freqs=self.J)
SCF = fot_tst.SmoothCFTest(test_freqs, gwidth, alpha=self.alpha)
result = SCF.perform_test(test)
p_val = result['pvalue']
return p_val
def job_scf_gwopt(sample_source, tr, te, r):
"""SmoothCFTest. Optimize only the Gaussian width.
Fix the test frequencies"""
op_gwidth = {'max_iter': 200, 'gwidth_step_size': 0.1,
'batch_proportion': 1.0, 'tol_fun': 1e-3}
# optimize on the training set
rand_state = np.random.get_state()
np.random.seed(seed=r+92856)
T_randn = np.random.randn(J, sample_source.dim())
np.random.set_state(rand_state)
gwidth, info = tst.SmoothCFTest.optimize_gwidth(tr, T_randn, **op_gwidth)
scf_gwopt = tst.SmoothCFTest(T_randn, gwidth, alpha)
raise ValueError('Use job_scf_gwgrid instead')
return scf_gwopt.perform_test(te)
def TST_SCF(Fea, N1, alpha, is_train, test_freqs, gwidth, J = 1, seed = 15):
"""run SCF test."""
Fea = get_item(Fea,is_cuda)
tst_data = data.TSTData(Fea[0:N1,:], Fea[N1:,:])
h = 0
if is_train:
op = {'n_test_freqs': J, 'seed': seed, 'max_iter': 300,
'batch_proportion': 1.0, 'freqs_step_size': 0.1,
'gwidth_step_size': 0.01, 'tol_fun': 1e-4}
test_freqs, gwidth, info = tst.SmoothCFTest.optimize_freqs_width(tst_data, alpha, **op)
return test_freqs, gwidth
else:
scf_opt = tst.SmoothCFTest(test_freqs, gwidth, alpha=alpha)
test_result = scf_opt.perform_test(tst_data)
if test_result['h0_rejected']:
h = 1
return h
def job_scf_gwgrid(sample_source, tr, te, r, J):
rand_state = np.random.get_state()
np.random.seed(r+92856)
d = tr.dim()
T_randn = np.random.randn(J, d)
np.random.set_state(rand_state)
# grid search to determine the initial gwidth
mean_sd = tr.mean_std()
scales = 2.0**np.linspace(-4, 4, 20)
list_gwidth = np.hstack( (mean_sd*scales*(d**0.5), 2**np.linspace(-10, 10, 20) ))
list_gwidth.sort()
besti, powers = tst.SmoothCFTest.grid_search_gwidth(tr, T_randn,
list_gwidth, alpha)
# initialize with the best width from the grid search
best_width = list_gwidth[besti]
scf_gwgrid = tst.SmoothCFTest(T_randn, best_width, alpha)
return scf_gwgrid.perform_test(te)
def job_scf_opt10(prob_label, tr, te, r, ni, n):
"""SmoothCFTest with frequencies optimized."""
with util.ContextTimer() as t:
op = {
'n_test_freqs': J,
'max_iter': 200,
'freqs_step_size': 2.0,
'gwidth_step_size': 0.2,
'seed': r + 92856,
'tol_fun': 1e-4
}
test_freqs, gwidth, info = tst.SmoothCFTest.optimize_freqs_width(
tr, alpha, **op)
scf_opt = tst.SmoothCFTest(test_freqs, gwidth, alpha)
scf_opt_test = scf_opt.perform_test(te)
return {
#'test_method': scf_opt,
'test_result': scf_opt_test,
'time_secs': t.secs
}
def test(self, X, Y):
XY = self.preprocess(X, Y)
train, test = XY.split_tr_te(tr_proportion=self.split_ratio)
freqs = np.random.randn(self.J, XY.dim())
mean_sd = train.mean_std()
scales = 2.**np.linspace(-4, 4, 30)
list_gwidth = np.hstack(
[mean_sd * scales * (XY.dim()**0.5), 2**np.linspace(-8, 8, 20)])
list_gwidth.sort()
with contextlib.redirect_stdout(None):
best_i, powers = fot_tst.SmoothCFTest.grid_search_gwidth(
train, freqs, list_gwidth, self.alpha)
best_width = list_gwidth[best_i]
SCF = fot_tst.SmoothCFTest(freqs, best_width, self.alpha)
result = SCF.perform_test(test)
p_val = result['pvalue']
return p_val
