from dlmnn.data.get_img_data import get_dataset
from dlmnn.LMNN import lmnn
from dlmnn.LMNNSequential import sequental_lmnn
from dlmnn.helper.argparser import lmnn_argparser
import numpy as np
#%%
if __name__ == '__main__':
# Input arguments
args = lmnn_argparser()
print(args)
# Get some data
x_train, y_train, x_test, y_test = get_dataset(args.dataset)
input_shape=x_train.shape[1:]
# Constants
baseMapNum = 32
weight_decay = 1e-4
num_classes = len(np.unique(y_train))
# Normalize data
mean = np.mean(x_train,axis=(0,1,2,3))
std = np.std(x_train,axis=(0,1,2,3))
x_train = (x_train-mean)/(std+1e-7)
x_test = (x_test-mean)/(std+1e-7)
if args.model_type == 'softmax':
y_train = utils.to_categorical(y_train,num_classes)
#%%
from dlmnn import sequental_lmnn
from dlmnn.helper.argparser import lmnn_argparser
from dlmnn.helper.layers import InputLayer, Flatten, Conv2D, MaxPool2D, \
Dense, LeakyReLU
from dlmnn.data.get_img_data import get_dataset
#%%
if __name__ == '__main__':
# Get input arguments
args = lmnn_argparser()
print(args)
# Get some data
X_train, y_train, X_test, y_test = get_dataset('cifar10')
input_shape = X_train.shape[1:]
# Add layers to model
model = sequental_lmnn()
model.add(InputLayer, input_shape=input_shape)
model.add(Conv2D, filters=16, kernel_size=(3, 3), padding='same')
model.add(LeakyReLU, alpha=0.3)
model.add(MaxPool2D, pool_size=(2, 2))
model.add(Conv2D, filters=32, kernel_size=(3, 3), padding='same')
model.add(LeakyReLU, alpha=0.3)
model.add(MaxPool2D, pool_size=(2, 2))
model.add(Conv2D, filters=64, kernel_size=(3, 3), padding='same')
model.add(LeakyReLU, alpha=0.3)
model.add(MaxPool2D, pool_size=(2, 2))
model.add(Flatten)
if __name__ == '__main__':
# Allocate array for results
performance = np.zeros((5, ), dtype=np.float32)
pred1, pred2, pred3, pred4, pred5 = [ ], [ ], [ ], [ ], [ ]
# Get input arguments
args = argparser()
dataset = args['dataset']
k = args['k']
maxEpochs = args['e']
name = args['n']
margin = args['m']
mu = args['w']
# Get data
X_train, y_train, X_test, y_test = get_dataset(dataset)
input_shape = X_train.shape[1:]
n_class = len(np.unique(y_train))
print(70*'=')
print(dataset)
print(70*'=')
############################## MODEL 1 ################################
############################## KNN classifier #########################
pred = knnClassifier(X_test, X_train, y_train, k=k)
acc = np.mean(pred==y_test)
# Save results
pred1.append(pred)
performance[0] = acc