class_num = 2
feature_num = pima.shape[1] - 1
x_train = torch.from_numpy(pima[:, 0:feature_num]).float().cuda()
y_train = torch.from_numpy(pima[:, feature_num]).cuda()
# data_num个特征，需要data_num组编码器
enc_layer = []
x_train_min = x_train.min(0)[0]
x_train_max = x_train.max(0)[0]
print(x_train_min)
print(x_train_max)
enc_neuron_num = 512
per_class_neuron_num = 1
for i in range(feature_num):
enc_layer.append(
encoder.GaussianEncoder(x_train_min[i], x_train_max[i],
enc_neuron_num, 'cuda:0'))
# 共有class_num类，需要class_num个tempotron
dec_layer = []
for i in range(class_num * per_class_neuron_num):
dec_layer.append(
node.LIFNode(tau=15.0,
tau_s=15.0 / 4,
v_rest=0,
T=500,
N=enc_neuron_num * feature_num,
device='cuda:0')) # 由于是全连接，因此enc_layer的所有元素都与这个节点相连
W = torch.rand(
size=[class_num * per_class_neuron_num, enc_neuron_num *
feature_num]).cuda() # W[i]表示enc_layer与dec_layer[i]的连接权重
B = torch.rand(
iris = datasets.load_iris() # 字典，iris.data为特征，iris.target为类别
# print(iris['data'].shape) # [150,4]
# print(iris['target'].shape) # [150]
x_train = torch.from_numpy(iris['data']).float().cuda()
y_train = torch.from_numpy(iris['target']).cuda()
# 4个特征，需要4组编码器
enc_layer = []
x_train_min = x_train.min(0)[0]
x_train_max = x_train.max(0)[0]
print(x_train_min)
print(x_train_max)
enc_neuron_num = 4096
per_class_neuron_num = 16
# [0, per_class_neuron_num-1]对应第0类，[per_class_neuron_num, 2*per_class_neuron_num-1]对应第1类，[2*per_class_neuron_num, 3*per_class_neuron_num-1]对应第2类
for i in range(4):
enc_layer.append(encoder.GaussianEncoder(x_train_min[i], x_train_max[i], enc_neuron_num, 'cuda:' + sys.argv[2]))
# 共有3类，需要3个tempotron
dec_layer = []
for i in range(3*per_class_neuron_num):
dec_layer.append(node.LIFNode(tau=15.0, tau_s=15.0/4, v_rest=0, T=500, N=enc_neuron_num*4, device='cuda:' + sys.argv[2])) # 由于是全连接，因此enc_layer的所有元素都与这个节点相连
W = torch.rand(size=[3*per_class_neuron_num, enc_neuron_num*4]).cuda() # W[i]表示enc_layer与dec_layer[i]的连接权重
B = torch.rand(size=[3*per_class_neuron_num, enc_neuron_num*4]).cuda()
learn_rate = 0.1
stdp_learn_rate = 1
A_LTD = 0.02656
C_ = 0.0001
A_LTP = C_ * A_LTD
tau_LTP = 16.8
tau_LTD = 33.7
