if __name__ == "__main__":
# kernels = [ utils.DoGKernel(3,1,2), utils.DoGKernel(3,2,1),
# utils.OnCenter(3), utils.OffCenter(3)]
kernels = [utils.DoGKernel(3,1,2), utils.DoGKernel(3,2,1)]
filter = utils.Filter(kernels, padding = 6, thresholds = 50)
transform = InputTransform(filter)
data_root = 'data/'
MNIST_train = utils.CacheDataset(MNIST(root=data_root, train=True, download=True, transform=transform)) # 60000 x 30 x 30
MNIST_test = utils.CacheDataset(MNIST(root=data_root, train=True, download=True, transform=transform)) # 10000 x 30
MNIST_loader = DataLoader(MNIST_train, batch_size=1000, shuffle=True)
MNIST_test_loader = DataLoader(MNIST_test, batch_size=1000, shuffle=False)
net = CTNN()
clf = svm.SVC()
net = train(net, MNIST_loader)
torch.save(net.state_dict(), "./MNISTcheckpoint.pt")
net.state_dict(torch.load("./MNISTcheckpoint.pt"))
train_outputs, train_y = inference(net, MNIST_loader)
test_outputs, test_y = inference(net, MNIST_test_loader)
train_outputs, test_outputs = preprocess(train_outputs, test_outputs)
kernels = [
utils.DoGKernel(3, 3 / 9, 6 / 9),
utils.DoGKernel(3, 6 / 9, 3 / 9),
utils.DoGKernel(7, 7 / 9, 14 / 9),
utils.DoGKernel(7, 14 / 9, 7 / 9),
utils.DoGKernel(13, 13 / 9, 26 / 9),
utils.DoGKernel(13, 26 / 9, 13 / 9)
]
filter = utils.Filter(kernels, padding=6, thresholds=50)
s1c1 = S1C1Transform(filter)
data_root = "data"
MNIST_train = utils.CacheDataset(
torchvision.datasets.MNIST(root=data_root,
train=True,
download=True,
transform=s1c1))
MNIST_test = utils.CacheDataset(
torchvision.datasets.MNIST(root=data_root,
train=False,
download=True,
transform=s1c1))
MNIST_loader = DataLoader(MNIST_train, batch_size=1000, shuffle=False)
MNIST_testLoader = DataLoader(MNIST_test,
batch_size=len(MNIST_test),
shuffle=False)
mozafari = MozafariMNIST2018()
if use_cuda:
mozafari.cuda()
utils.GaborKernel(5, 45 + 22.5),
utils.GaborKernel(5, 90 + 22.5),
utils.GaborKernel(5, 135 + 22.5),
utils.GaborKernel(5, 180 + 22.5)
]
filter = utils.Filter(kernels, use_abs=True)
lateral_inhibition = utils.LateralIntencityInhibition(
[0.15, 0.12, 0.1, 0.07, 0.05])
task = "Caltech"
use_cuda = True
if task == "Caltech":
s1c1 = S1C1Transform(filter, 7, 6, lateral_inhibition)
trainsetfolder = utils.CacheDataset(ImageFolder("facemotortrain", s1c1))
testsetfolder = utils.CacheDataset(ImageFolder("facemotortest", s1c1))
mozafari = Mozafari2018(4, 10, 2, (17, 17), 42, (0.005, -0.0025),
(-0.005, 0.0005), 0.5)
trainset = DataLoader(trainsetfolder,
batch_size=len(trainsetfolder),
shuffle=True)
testset = DataLoader(testsetfolder,
batch_size=len(testsetfolder),
shuffle=True)
max_epoch = 400
elif task == "ETH":
s1c1 = S1C1Transform(filter, 5, 4, lateral_inhibition)
mozafari = Mozafari2018(4, 10, 8, (31, 31), 160, (0.01, -0.0035),
(-0.01, 0.0006), 0.4)
dataset = ImageFolder("D:\PyCharm 2020.3.5\pythonProject\dataset\eth",
transform) # adding transform to the dataset
# splitting training and testing sets
indices = list(range(len(dataset)))
random.shuffle(indices)
split_point = int(0.75 * len(indices))
train_indices = indices[:split_point]
test_indices = indices[split_point:]
print("Size of the training set:", len(train_indices))
print("Size of the testing set:", len(test_indices))
from torch.utils.data import DataLoader
from torch.utils.data import SubsetRandomSampler
dataset = utils.CacheDataset(dataset)
train_loader = DataLoader(dataset, sampler=SubsetRandomSampler(train_indices))
test_loader = DataLoader(dataset, sampler=SubsetRandomSampler(test_indices))
import SpykeTorch.snn as snn
pool = snn.Pooling(kernel_size=3, stride=2)
conv = snn.Convolution(in_channels=4, out_channels=20, kernel_size=30)
stdp = snn.STDP(conv_layer=conv, learning_rate=(0.05, -0.015))
for iter in range(300):
print('\rIteration:', iter, end="")
for data, _ in train_loader:
for x in data:
x = pool(x)
p = conv(x)
o, p = sf.fire(p, 20, return_thresholded_potentials=True)
temporal_image = self.temporal_transform(image)
return temporal_image.sign().byte()
kernels = [
utils.DoGKernel(7, 1, 2),
utils.DoGKernel(7, 2, 1),
]
filter = utils.Filter(kernels, padding=3, thresholds=50, type="gray")
s1 = S1Transform(filter)
data_root = "data"
# MNIST_train = utils.CacheDataset(torchvision.datasets.MNIST(root=data_root, train=True, download=True, transform=s1))
# print(type(MNIST_train))
MNIST_test = utils.CacheDataset(ImageFolder(root="demo", transform=s1))
# MNIST_test = utils.CacheDataset(torchvision.datasets.MNIST(root=data_root, train=False, download=True, transform=s1))
# MNIST_loader = DataLoader(MNIST_train, batch_size=len(MNIST_train), shuffle=False)
MNIST_testLoader = DataLoader(MNIST_test,
batch_size=len(MNIST_test),
shuffle=False)
kheradpisheh = KheradpishehMNIST()
if use_cuda:
kheradpisheh.cuda()
# Training The First Layer
print("load weights for the first layer")
if os.path.isfile("weights/saved_l1.net"):
if use_cuda:
kheradpisheh.load_state_dict(torch.load("weights/saved_l1.net"))
xtest = sample_zero_mean(xtest)
xtest = gcn(xtest)
return x, xtest
if __name__ == "__main__":
kernels = [ utils.DoGKernel(3,1,2), utils.DoGKernel(3,2,1)]
filter = utils.Filter(kernels, padding = 6, thresholds = 50)
transform = InputTransform(filter)
data_root = 'data/'
cifar_data_root = 'cifar/'
MNIST_train = utils.CacheDataset(MNIST(root=data_root, train=True, download=True, transform=transform)) # 60000 x 28 x 28
MNIST_test = utils.CacheDataset(MNIST(root=data_root, train=False, download=True, transform=transform)) # 10000 x 28
MNIST_loader = DataLoader(MNIST_train, batch_size=1000, shuffle=True)
MNIST_test_loader = DataLoader(MNIST_test, batch_size=1000, shuffle=False)
CIFAR_train = utils.CacheDataset(CIFAR(root=cifar_data_root, train=True, download=True, transform=transform)) # 60000 x 30 x 30
CIFAR_test = utils.CacheDataset(CIFAR(root=cifar_data_root, train=False, download=True, transform=transform)) # 10000 x 30
CIFAR_loader = DataLoader(CIFAR_train, batch_size=1000, shuffle=True)
CIFAR_test_loader = DataLoader(CIFAR_test, batch_size=1000, shuffle=False)
net = CTNN()
clf = svm.SVC(verbose=True) # (60000 x 100)