def setUp(self):
np.random.seed(SEED)
self.G_params = {}
self.G_params['type'] = ds.SBM
self.G_params['N'] = 32
self.G_params['k'] = 4
self.G_params['p'] = 0.8
self.G_params['q'] = 0.1
self.G_params['type_z'] = ds.RAND
self.dest = 5
self.Gx, self.Gy = ds.nodes_perturbated_graphs(self.G_params, self.dest,
seed=SEED)
def test_permute_graph(self):
dest = 30
Gx, Gy = ds.nodes_perturbated_graphs(self.G_params, dest,
perm=True, seed=SEED)
Ax = Gx.W.todense()
Ay = Gy.W.todense()
P = Gy.info['perm_matrix']
Ax_rm = np.delete(Ax, Gy.info['rm_nodes'], axis=0)
Ax_rm = np.delete(Ax_rm, Gy.info['rm_nodes'], axis=1)
rm_comms_X = np.delete(Gx.info['node_com'], Gy.info['rm_nodes'])
comm_Y = Gy.info['node_com']
self.assertFalse(np.array_equal(Ax_rm, Ay))
self.assertFalse(np.array_equal(rm_comms_X, comm_Y))
self.assertTrue(np.array_equal(np.eye(Gy.N), P.dot(P.T)))
self.assertTrue(np.array_equal(Ax_rm, P.T.dot(Ay).dot(P)))
self.assertTrue(np.array_equal(rm_comms_X, P.T.dot(comm_Y)))
def test_model(id, signals, nn_params, model_params):
Gx, Gy = data_sets.nodes_perturbated_graphs(signals['g_params'],
signals['pert'],
perm=signals['perm'])
# Define the data model
data = data_sets.LinearDS2GSNodesPert(
Gx,
Gy,
signals['N_samples'],
signals['L_filter'],
signals['g_params']['k'], # k is n_delts
median=signals['median'])
data.to_unit_norm()
data.add_noise(signals['noise'], test_only=signals['test_only'])
data.to_tensor()
if nn_params['arch_type'] == "basic":
Gx.compute_laplacian('normalized')
archit = BasicArch(Gx.L.todense(), nn_params['F'], nn_params['K'],
nn_params['M'], nn_params['nonlin'], ARCH_INFO)
elif nn_params['arch_type'] == "mlp":
archit = MLP(nn_params['F'], nn_params['nonlin'], ARCH_INFO)
elif nn_params['arch_type'] == "conv":
archit = ConvNN(N, nn_params['F'], nn_params['K'], nn_params['nonlin'],
nn_params['M'], ARCH_INFO)
elif nn_params['arch_type'] == "linear":
archit = MLP(nn_params['F'], nn_params['nonlin'], ARCH_INFO)
else:
raise RuntimeError("arch_type has to be either basic, mlp or conv")
model_params['arch'] = archit
model = Model(**model_params)
t_init = time.time()
epochs, _, _ = model.fit(data.train_X, data.train_Y, data.val_X,
data.val_Y)
t_conv = time.time() - t_init
mean_err, med_err, mse = model.test(data.test_X, data.test_Y)
print(
"DONE {}: MSE={} - Mean Err={} - Median Err={} - Params={} - t_conv={} - epochs={}"
.format(id, mse, mean_err, med_err, model.count_params(),
round(t_conv, 4), epochs),
flush=True)
return mse, mean_err, med_err, model.count_params(), t_conv, epochs
def test_permutated_S(self):
n_samps = [50, 20, 20]
L = 6
n_delts = self.G_params['k']
Gx, Gy = ds.nodes_perturbated_graphs(self.G_params, self.dest,
perm=True, seed=SEED)
data = ds.LinearDS2GSNodesPert(Gx, Gy, n_samps, L, n_delts)
P = data.Gy.info['perm_matrix']
rm_nodes = Gy.info['rm_nodes']
train_Sx_rm = np.delete(data.train_Sx, rm_nodes, axis=1)
val_Sx_rm = np.delete(data.val_Sx, rm_nodes, axis=1)
test_Sx_rm = np.delete(data.test_Sx, rm_nodes, axis=1)
self.assertFalse(np.array_equal(train_Sx_rm, data.train_Sy))
self.assertFalse(np.array_equal(val_Sx_rm, data.val_Sy))
self.assertFalse(np.array_equal(test_Sx_rm, data.test_Sy))
self.assertTrue(np.array_equal(train_Sx_rm, data.train_Sy.dot(P)))
self.assertTrue(np.array_equal(val_Sx_rm, data.val_Sy.dot(P)))
self.assertTrue(np.array_equal(test_Sx_rm, data.test_Sy.dot(P)))
def test_perturbation(self):
dest = 30
for i in range(10):
Gx, Gy = ds.nodes_perturbated_graphs(self.G_params, dest,
seed=SEED)
Ax = Gx.W.todense()
Ay = Gy.W.todense()
Ax_rm = np.delete(Ax, Gy.info['rm_nodes'], axis=0)
Ax_rm = np.delete(Ax_rm, Gy.info['rm_nodes'], axis=1)
rm_comms = np.delete(Gx.info['node_com'], Gy.info['rm_nodes'])
self.assertFalse(Gx.is_directed())
self.assertTrue(Gx.is_connected())
self.assertEqual(np.sum(np.diag(Ax)), 0)
self.assertFalse(Gy.is_directed())
self.assertTrue(Gy.is_connected())
self.assertEqual(np.sum(np.diag(Ay)), 0)
self.assertEqual(Gy.N, Gx.N-dest)
self.assertTrue(np.array_equal(Ax_rm, Ay))
self.assertTrue(np.array_equal(rm_comms, Gy.info['node_com']))
def test_model(signals, nn_params, model_params):
Gx, Gy = data_sets.nodes_perturbated_graphs(signals['g_params'],
signals['pert'],
perm=signals['perm'],
seed=SEED)
# Define the data model
data = data_sets.LinearDS2GSNodesPert(
Gx,
Gy,
signals['N_samples'],
signals['L_filter'],
signals['g_params']['k'], # k is n_delts
median=signals['median'])
data.to_unit_norm()
data.add_noise(signals['noise'], test_only=signals['test_only'])
data.to_tensor()
Gx.compute_laplacian('normalized')
Gy.compute_laplacian('normalized')
archit = GIGOArch(Gx.L.todense(), Gy.L.todense(), nn_params['Fi'],
nn_params['Fo'], nn_params['Ki'], nn_params['Ko'],
nn_params['C'], nn_params['nonlin'],
nn_params['last_act_fn'], nn_params['batch_norm'],
nn_params['arch_info'])
model_params['arch'] = archit
model = Model(**model_params)
t_init = time.time()
epochs, _, _ = model.fit(data.train_X, data.train_Y, data.val_X,
data.val_Y)
t_conv = time.time() - t_init
mean_err, med_err, mse = model.test(data.test_X, data.test_Y)
print(
"DONE: MSE={} - Mean Err={} - Median Err={} - Params={} - t_conv={} - epochs={}"
.format(mse, mean_err, med_err, model.count_params(), round(t_conv, 4),
epochs))
return mse, med_err, mean_err, model.count_params(), t_conv, epochs
def run(id, Gs, Signals, lrn, pert):
if Gs['params']['type'] == ds.SBM:
Gx, Gy = ds.nodes_perturbated_graphs(Gs['params'],
pert,
seed=SEED,
perm=True)
elif Gs['params']['type'] == ds.BA:
Gx = ds.create_graph(Gs['params'], SEED)
G_params_y = Gs['params'].copy()
G_params_y['N'] = Gs['params']['N'] - pert
Gy = ds.create_graph(G_params_y, 2 * SEED)
else:
raise RuntimeError("Choose a valid graph type")
data = ds.LinearDS2GSNodesPert(Gx,
Gy,
Signals['samples'],
Signals['L'],
Signals['deltas'],
median=Signals['median'],
same_coeffs=Signals['same_coeffs'],
neg_coeffs=Signals['neg_coeffs'])
data.to_unit_norm()
data.add_noise(Signals['noise'], test_only=Signals['test_only'])
data.to_tensor()
epochs = 0
params = np.zeros(N_EXPS)
med_err = np.zeros(N_EXPS)
mse = np.zeros(N_EXPS)
for i, exp in enumerate(EXPS):
if exp['type'] == 'Linear':
model = LinearModel(exp['N'])
elif exp['type'] == 'Enc_Dec':
exp['n_dec'][-1] = Gy.N
clust_x = gc.MultiResGraphClustering(Gx,
exp['n_enc'],
k=exp['n_enc'][-1],
up_method=exp['downs'])
clust_y = gc.MultiResGraphClustering(Gy,
exp['n_dec'],
k=exp['n_enc'][-1],
up_method=exp['ups'])
net = GraphEncoderDecoder(exp['f_enc'],
clust_x.sizes,
clust_x.Ds,
exp['f_dec'],
clust_y.sizes,
clust_y.Us,
exp['f_conv'],
As_dec=clust_y.As,
K_dec=exp['K_dec'],
K_enc=exp['K_enc'],
As_enc=clust_x.As,
act_fn=lrn['af'],
last_act_fn=lrn['laf'],
ups=exp['ups'],
downs=exp['downs'])
elif exp['type'] == 'AutoConv':
conv = exp['convs'][PERT.index(pert)]
net = ConvAutoencoder(conv['f_enc'], conv['kernel_enc'],
conv['f_dec'], conv['kernel_dec'])
elif exp['type'] == 'AutoFC':
exp['n_dec'][-1] = Gy.N
net = FCAutoencoder(exp['n_enc'], exp['n_dec'], bias=exp['bias'])
else:
raise RuntimeError('Unknown experiment type')
if exp['type'] != 'Linear':
model = Model(net,
learning_rate=lrn['lr'],
decay_rate=lrn['dr'],
batch_size=lrn['batch'],
epochs=lrn['epochs'],
eval_freq=EVAL_F,
max_non_dec=lrn['non_dec'],
verbose=VERBOSE,
early_stop=exp['early_stop'])
epochs, _, _ = model.fit(data.train_X, data.train_Y, data.val_X,
data.val_Y)
_, med_err[i], mse[i] = model.test(data.test_X, data.test_Y)
params[i] = model.count_params()
print('G: {}, {}-{} ({}): epochs {} - mse {} - MedianErr: {}'.format(
id, i, exp['type'], params[i], epochs, mse[i], med_err[i]))
return params, med_err, mse