def formatit(X0, X1, Y0, Y1):
w = list(range(int(np.ceil(X0)), int(X1) + 1))
plt.xticks(w, [str(t) for t in w])
w = list(range(int(np.ceil(Y0)), int(Y1) + 1))
plt.yticks(w, [str(t) for t in w])
for ss in [0.5, 0.2, 0.5]:
np.random.seed(int(sys.argv[-1]) + 10)
sigma_x = np.eye(Ds[-1]) * ss
for i in range(5):
fig = plt.figure(figsize=(5, 5))
model = networks.create_fns(BS, R, Ds, 0, var_x=np.ones(2) * ss**2)
output, A, b, inequalities, signs = model['input2all'](np.random.randn(
Ds[0]))
regions = utils.search_region(model['signs2ineq'], model['signs2Ab'],
signs)
As = np.array([regions[s]['Ab'][0] for s in regions])
Bs = np.array([regions[s]['Ab'][1] for s in regions])
predictions = np.array([
model['input2all'](np.random.randn(Ds[0]))[0] for z in range(200)
])
noise = np.random.randn(*predictions.shape) * ss
plt.imshow(predictions[i].reshape((8, 8)))
for i in range(50):
plt.subplot(10, 10, 51 + i)
plt.imshow(DATA[i].reshape((8, 8)))
np.random.seed(int(sys.argv[-1]) + 10)
Ds = [1, 12, 64]
R, BS = 110, 150
sigma_x = .1
model = networks.create_fns(BS, R, Ds, 1, lr=0.001, leakiness=0.1, var_x=sigma_x**2)
DATA = datasets.load_digits(n_class=1).images
DATA = DATA[:BS].reshape((BS, -1)) + np.random.randn(BS, Ds[-1]) * 0.1
#DATA -= DATA.mean(1, keepdims=True)
DATA /= (DATA.max() * 10)
print(DATA)
L = []
for iter in tqdm(range(36)):
plt_state()
plt.savefig('samples_{}.png'.format(iter))
plt.close()
L.append(networks.EM(model, DATA, 1, 10, pretrain=iter==0, update_var=iter>5))
import sys
sys.path.insert(0, "../../SymJAX")
sys.path.insert(0, "../")
import numpy as np
import symjax as sj
import symjax.tensor as T
import utils
import networks
from tqdm import tqdm
error1, error2 = [], []
for Ds in [[1, 8, 8]]:
print(Ds)
model = networks.create_fns(1, 1, Ds, 0, var_x=np.ones(Ds[-1]))
x = np.random.randn(Ds[0])/10
output, A, b, inequalities, signs = model['input2all'](x)
regions = utils.search_region(model['signs2ineq'], model['signs2Ab'],
signs)
print(len(regions))
K = 100
xx = [np.linspace(-10, 10, K)] * Ds[0]
xx = np.meshgrid(*xx)
xx = np.vstack([x.flatten() for x in xx]).T
yy = np.zeros((int(K ** Ds[0]), Ds[-1]))
yy2 = np.zeros((int(K ** Ds[0]), Ds[-1]))
As = np.array([regions[s]['Ab'][0] for s in regions])
Bs = np.array([regions[s]['Ab'][1] for s in regions])
def plt_state():
predictions = model['sample'](100)
plt.figure(figsize=(30, 30))
for i in range(50):
plt.subplot(10, 10, 1 + i)
plt.imshow(predictions[i].reshape((8, 8)))
for i in range(50):
plt.subplot(10, 10, 51 + i)
plt.imshow(DATA[i].reshape((8, 8)))
np.random.seed(int(sys.argv[-1]) + 10)
Ds = [1, 20, 64]
R, BS = 110, 150
model = networks.create_fns(BS, R, Ds, 0, var_x=np.ones(Ds[-1]), lr=0.001)
DATA = datasets.load_digits(n_class=1).images
DATA = DATA[:BS].reshape((BS, -1))
DATA /= (0.1 + DATA.max(1, keepdims=True))
DATA -= DATA.mean(1, keepdims=True)
DATA += np.random.randn(BS, Ds[-1]) * 0.3
L = []
for iter in tqdm(range(16)):
L.append(networks.EM(model, DATA, 100))
plt_state()
plt.savefig('samples_{}.png'.format(iter))
plt.close()
plt.figure()
if Ds[0] == 1:
xx = np.linspace(-5, 5, 1500).reshape((-1, 1))
else:
xx = np.meshgrid(np.linspace(-5, 5, 100), np.linspace(-3, 3, 100))
xx = np.vstack([xx[0].flatten(), xx[1].flatten()]).T
ss = 0.1
for seed in [37, 146, 53, 187, 79]:
np.random.seed(seed)
for i in range(1):
fig = plt.figure(figsize=(5, 5))
model = networks.create_fns(BS,
R,
Ds,
seed,
var_x=np.ones(Ds[-1]) * ss**2,
leakiness=0.01)
z = np.random.randn(Ds[0])
output, A, b, inequalities, signs = model['input2all'](z)
outpute = output + np.random.randn(Ds[-1]) * ss
regions = utils.search_region(model['signs2ineq'], model['signs2Ab'],
signs)
print(len(regions))
As = np.array([regions[s]['Ab'][0] for s in regions])
Bs = np.array([regions[s]['Ab'][1] for s in regions])
predictions = model['sample'](20000)
elif args.network == 'large':
Ds = [1, 8, 8, DATA.shape[1]]
R = 64
elif args.network == 'xlarge':
Ds = [1, 32, 32, DATA.shape[1]]
R = 128
graph = sj.Graph('test')
with graph:
if args.model != 'EM':
lr = sj.tensor.Variable(1., name='lr', trainable=False)
emt = networks.create_fns(BS,
R,
Ds,
1,
var_x=np.ones(Ds[-1]) * args.std**2,
lr=0.005,
leakiness=args.leakiness)
model = networks.create_vae(50,
Ds,
args.seed,
lr=lr,
leakiness=args.leakiness,
scaler=1)
else:
model = networks.create_fns(BS,
R,
Ds,
1,
lr=0.001,
parser.add_argument('--pretrain', type=int)
parser.add_argument('--seed', type=int)
args = parser.parse_args()
np.random.seed(args.seed)
Ds = [1, 12, 2]
R, BS = 64, 500
#if sys.argv[-1] == 'VAE':
# model = networks.create_vae(BS, Ds, 1, lr=0.005, leakiness=0.1)
#else:
model = networks.create_fns(BS,
R,
Ds,
1,
lr=0.001,
leakiness=0.1,
var_x=args.std**2)
X = model['sample'](BS)
noise = np.random.randn(*X.shape) * np.sqrt(model['varx']())
plt.scatter(X[:, 0], X[:, 1])
plt.scatter(X[:, 0] + noise[:, 0], X[:, 1] + noise[:, 1])
plt.savefig('ori.png')
plt.close()
DATA = np.random.randn(BS, Ds[-1])
DATA /= np.linalg.norm(DATA, 2, 1, keepdims=True)
DATA = np.random.randn(BS) * 2