def wider_pre_conv(layer, n_add_filters, weighted=True):
n_dim = get_n_dim(layer)
if not weighted:
return get_conv_class(n_dim)(layer.input_channel,
layer.filters + n_add_filters,
kernel_size=layer.kernel_size,
stride=layer.stride,
padding =layer.padding,
groups=layer.groups)
n_pre_filters = layer.filters
rand = np.random.randint(n_pre_filters, size=n_add_filters)
teacher_w, teacher_b = layer.get_weights()
student_w = teacher_w.copy()
student_b = teacher_b.copy()
# target layer update (i)
for i in range(len(rand)):
teacher_index = rand[i]
new_weight = teacher_w[teacher_index, ...]
new_weight = new_weight[np.newaxis, ...]
student_w = np.concatenate((student_w, new_weight), axis=0)
student_b = np.append(student_b, teacher_b[teacher_index])
new_pre_layer = get_conv_class(n_dim)(layer.input_channel,
n_pre_filters + n_add_filters,
kernel_size=layer.kernel_size,
stride=layer.stride,
padding =layer.padding,
groups=layer.groups)
new_pre_layer.set_weights((add_noise(student_w, teacher_w), add_noise(student_b, teacher_b)))
return new_pre_layer
def wider_next_conv(layer, start_dim, total_dim, n_add, weighted=True):
n_dim = get_n_dim(layer)
groups = layer.groups + n_add if layer.groups is not None and layer.groups > 1 else 1
if not weighted:
return get_conv_class(n_dim)(layer.input_channel + n_add,
layer.filters,
kernel_size=layer.kernel_size,
stride=layer.stride,
padding =layer.padding,
groups =groups)
n_filters = layer.filters
new_layer = get_conv_class(n_dim)(layer.input_channel + n_add,
n_filters,
kernel_size=layer.kernel_size,
stride=layer.stride,
padding =layer.padding,
groups =groups)
teacher_w, teacher_b = layer.get_weights()
if layer.groups is not None and layer.groups > 1:
new_layer.set_weights((teacher_w, teacher_b))
else:
new_weight_shape = list(teacher_w.shape)
new_weight_shape[1] = n_add
new_weight = np.zeros(tuple(new_weight_shape))
student_w = np.concatenate((teacher_w[:, :start_dim, ...].copy(),
add_noise(new_weight, teacher_w),
teacher_w[:, start_dim:total_dim, ...].copy()), axis=1)
new_layer.set_weights((student_w, teacher_b))
return new_layer
def wider_pre_conv(layer, n_add_filters, weighted=True):
n_dim = get_n_dim(layer)
if not weighted:
return get_conv_class(n_dim)(layer.input_channel,
layer.filters + n_add_filters,
kernel_size=layer.kernel_size,
stride=layer.stride)
n_pre_filters = layer.filters
rand = np.random.randint(n_pre_filters, size=n_add_filters)
teacher_w, teacher_b = layer.get_weights()
student_w = teacher_w.copy()
student_b = teacher_b.copy()
# target layer update (i)
for i in range(len(rand)):
teacher_index = rand[i]
new_weight = teacher_w[teacher_index, ...]
new_weight = new_weight[np.newaxis, ...]
student_w = np.concatenate((student_w, new_weight), axis=0)
student_b = np.append(student_b, teacher_b[teacher_index])
new_pre_layer = get_conv_class(n_dim)(layer.input_channel,
n_pre_filters + n_add_filters,
kernel_size=layer.kernel_size,
stride=layer.stride)
new_pre_layer.set_weights((add_noise(student_w, teacher_w), add_noise(student_b, teacher_b)))
return new_pre_layer
def deeper_conv_block(conv_layer, kernel_size, weighted=True):
n_dim = get_n_dim(conv_layer)
filter_shape = (kernel_size, ) * 2
n_filters = conv_layer.filters
weight = np.zeros((n_filters, n_filters) + filter_shape)
center = tuple(map(lambda x: int((x - 1) / 2), filter_shape))
for i in range(n_filters):
filter_weight = np.zeros((n_filters, ) + filter_shape)
index = (i, ) + center
filter_weight[index] = 1
weight[i, ...] = filter_weight
bias = np.zeros(n_filters)
new_conv_layer = get_conv_class(n_dim)(conv_layer.filters,
n_filters,
kernel_size=kernel_size)
bn = get_batch_norm_class(n_dim)(n_filters)
if weighted:
new_conv_layer.set_weights(
(add_noise(weight,
np.array([0, 1])), add_noise(bias, np.array([0, 1]))))
new_weights = [
add_noise(np.ones(n_filters, dtype=np.float32), np.array([0, 1])),
add_noise(np.zeros(n_filters, dtype=np.float32), np.array([0, 1])),
add_noise(np.zeros(n_filters, dtype=np.float32), np.array([0, 1])),
add_noise(np.ones(n_filters, dtype=np.float32), np.array([0, 1]))
]
bn.set_weights(new_weights)
return [StubReLU(), new_conv_layer, bn]
def init_conv_weight(layer):
n_filters = layer.filters
filter_shape = (layer.kernel_size, ) * get_n_dim(layer)
weight = np.zeros((n_filters, n_filters) + filter_shape)
center = tuple(map(lambda x: int((x - 1) / 2), filter_shape))
for i in range(n_filters):
filter_weight = np.zeros((n_filters,) + filter_shape)
index = (i,) + center
filter_weight[index] = 1
weight[i, ...] = filter_weight
bias = np.zeros(n_filters)
layer.set_weights((add_noise(weight, np.array([0, 1])), add_noise(bias, np.array([0, 1]))))
def init_conv_weight(layer):
n_filters = layer.filters
filter_shape = (layer.kernel_size, ) * get_n_dim(layer)
weight = np.zeros((n_filters, n_filters) + filter_shape)
center = tuple(map(lambda x: int((x - 1) / 2), filter_shape))
for i in range(n_filters):
filter_weight = np.zeros((n_filters,) + filter_shape)
index = (i,) + center
filter_weight[index] = 1
weight[i, ...] = filter_weight
bias = np.zeros(n_filters)
layer.set_weights((add_noise(weight, np.array([0, 1])), add_noise(bias, np.array([0, 1]))))
def wider_bn(layer, start_dim, total_dim, n_add, weighted=True):
n_dim = get_n_dim(layer)
if not weighted:
return get_batch_norm_class(n_dim)(layer.num_features + n_add)
weights = layer.get_weights()
new_weights = [add_noise(np.ones(n_add, dtype=np.float32), np.array([0, 1])),
add_noise(np.zeros(n_add, dtype=np.float32), np.array([0, 1])),
add_noise(np.zeros(n_add, dtype=np.float32), np.array([0, 1])),
add_noise(np.ones(n_add, dtype=np.float32), np.array([0, 1]))]
student_w = tuple()
for weight, new_weight in zip(weights, new_weights):
temp_w = weight.copy()
temp_w = np.concatenate((temp_w[:start_dim], new_weight, temp_w[start_dim:total_dim]))
student_w += (temp_w,)
new_layer = get_batch_norm_class(n_dim)(layer.num_features + n_add)
new_layer.set_weights(student_w)
return new_layer
def wider_bn(layer, start_dim, total_dim, n_add, weighted=True):
n_dim = get_n_dim(layer)
if not weighted:
return get_batch_norm_class(n_dim)(layer.num_features + n_add)
weights = layer.get_weights()
new_weights = [add_noise(np.ones(n_add, dtype=np.float32), np.array([0, 1])),
add_noise(np.zeros(n_add, dtype=np.float32), np.array([0, 1])),
add_noise(np.zeros(n_add, dtype=np.float32), np.array([0, 1])),
add_noise(np.ones(n_add, dtype=np.float32), np.array([0, 1]))]
student_w = tuple()
for weight, new_weight in zip(weights, new_weights):
temp_w = weight.copy()
temp_w = np.concatenate((temp_w[:start_dim], new_weight, temp_w[start_dim:total_dim]))
student_w += (temp_w,)
new_layer = get_batch_norm_class(n_dim)(layer.num_features + n_add)
new_layer.set_weights(student_w)
return new_layer
def wider_next_conv(layer, start_dim, total_dim, n_add, weighted=True):
n_dim = get_n_dim(layer)
if not weighted:
return get_conv_class(n_dim)(layer.input_channel + n_add,
layer.filters,
kernel_size=layer.kernel_size)
n_filters = layer.filters
teacher_w, teacher_b = layer.get_weights()
new_weight_shape = list(teacher_w.shape)
new_weight_shape[1] = n_add
new_weight = np.zeros(tuple(new_weight_shape))
student_w = np.concatenate(
(teacher_w[:, :start_dim, ...].copy(), add_noise(
new_weight, teacher_w), teacher_w[:, start_dim:total_dim,
...].copy()),
axis=1)
new_layer = get_conv_class(n_dim)(layer.input_channel + n_add, n_filters,
layer.kernel_size)
new_layer.set_weights((student_w, teacher_b))
return new_layer
def wider_next_conv(layer, start_dim, total_dim, n_add, weighted=True):
n_dim = get_n_dim(layer)
if not weighted:
return get_conv_class(n_dim)(layer.input_channel + n_add,
layer.filters,
kernel_size=layer.kernel_size,
stride=layer.stride)
n_filters = layer.filters
teacher_w, teacher_b = layer.get_weights()
new_weight_shape = list(teacher_w.shape)
new_weight_shape[1] = n_add
new_weight = np.zeros(tuple(new_weight_shape))
student_w = np.concatenate((teacher_w[:, :start_dim, ...].copy(),
add_noise(new_weight, teacher_w),
teacher_w[:, start_dim:total_dim, ...].copy()), axis=1)
new_layer = get_conv_class(n_dim)(layer.input_channel + n_add,
n_filters,
kernel_size=layer.kernel_size,
stride=layer.stride)
new_layer.set_weights((student_w, teacher_b))
return new_layer
