474 lines
15 KiB
Python
474 lines
15 KiB
Python
|
import torch
|
||
|
|
|
||
|
|
from network.calculate_output_size import calculate_output_size
|
||
|
|
|
||
|
|
|
||
|
|
class NNMFLayerSbSBP(torch.nn.Module):
|
||
|
|
|
||
|
|
_epsilon_0: float
|
||
|
|
_weights: torch.nn.parameter.Parameter
|
||
|
|
_weights_exists: bool = False
|
||
|
|
_kernel_size: list[int]
|
||
|
|
_stride: list[int]
|
||
|
|
_dilation: list[int]
|
||
|
|
_padding: list[int]
|
||
|
|
_output_size: torch.Tensor
|
||
|
|
_number_of_neurons: int
|
||
|
|
_number_of_input_neurons: int
|
||
|
|
_h_initial: torch.Tensor | None = None
|
||
|
|
_w_trainable: bool
|
||
|
|
_weight_noise_range: list[float]
|
||
|
|
_input_size: list[int]
|
||
|
|
_output_layer: bool = False
|
||
|
|
_number_of_iterations: int
|
||
|
|
_local_learning: bool = False
|
||
|
|
|
||
|
|
device: torch.device
|
||
|
|
default_dtype: torch.dtype
|
||
|
|
|
||
|
|
_number_of_grad_weight_contributions: float = 0.0
|
||
|
|
|
||
|
|
last_input_store: bool = False
|
||
|
|
last_input_data: torch.Tensor | None = None
|
||
|
|
|
||
|
|
_layer_id: int = -1
|
||
|
|
|
||
|
|
_skip_gradient_calculation: bool
|
||
|
|
|
||
|
|
_keep_last_grad_scale: bool
|
||
|
|
_disable_scale_grade: bool
|
||
|
|
_last_grad_scale: torch.nn.parameter.Parameter
|
||
|
|
|
||
|
|
def __init__(
|
||
|
|
self,
|
||
|
|
number_of_input_neurons: int,
|
||
|
|
number_of_neurons: int,
|
||
|
|
input_size: list[int],
|
||
|
|
forward_kernel_size: list[int],
|
||
|
|
number_of_iterations: int,
|
||
|
|
epsilon_0: float = 1.0,
|
||
|
|
weight_noise_range: list[float] = [0.0, 1.0],
|
||
|
|
strides: list[int] = [1, 1],
|
||
|
|
dilation: list[int] = [0, 0],
|
||
|
|
padding: list[int] = [0, 0],
|
||
|
|
w_trainable: bool = False,
|
||
|
|
device: torch.device | None = None,
|
||
|
|
default_dtype: torch.dtype | None = None,
|
||
|
|
layer_id: int = -1,
|
||
|
|
local_learning: bool = False,
|
||
|
|
output_layer: bool = False,
|
||
|
|
skip_gradient_calculation: bool = False,
|
||
|
|
keep_last_grad_scale: bool = False,
|
||
|
|
disable_scale_grade: bool = True,
|
||
|
|
) -> None:
|
||
|
|
super().__init__()
|
||
|
|
|
||
|
|
assert device is not None
|
||
|
|
assert default_dtype is not None
|
||
|
|
self.device = device
|
||
|
|
self.default_dtype = default_dtype
|
||
|
|
|
||
|
|
self._w_trainable = bool(w_trainable)
|
||
|
|
self._stride = strides
|
||
|
|
self._dilation = dilation
|
||
|
|
self._padding = padding
|
||
|
|
self._kernel_size = forward_kernel_size
|
||
|
|
self._number_of_input_neurons = int(number_of_input_neurons)
|
||
|
|
self._number_of_neurons = int(number_of_neurons)
|
||
|
|
self._epsilon_0 = float(epsilon_0)
|
||
|
|
self._number_of_iterations = int(number_of_iterations)
|
||
|
|
self._weight_noise_range = weight_noise_range
|
||
|
|
self._layer_id = layer_id
|
||
|
|
self._local_learning = local_learning
|
||
|
|
self._output_layer = output_layer
|
||
|
|
self._skip_gradient_calculation = skip_gradient_calculation
|
||
|
|
|
||
|
|
self._keep_last_grad_scale = bool(keep_last_grad_scale)
|
||
|
|
self._disable_scale_grade = bool(disable_scale_grade)
|
||
|
|
self._last_grad_scale = torch.nn.parameter.Parameter(
|
||
|
|
torch.tensor(-1.0, dtype=self.default_dtype),
|
||
|
|
requires_grad=True,
|
||
|
|
)
|
||
|
|
|
||
|
|
assert len(input_size) == 2
|
||
|
|
self._input_size = input_size
|
||
|
|
|
||
|
|
self._output_size = calculate_output_size(
|
||
|
|
value=input_size,
|
||
|
|
kernel_size=self._kernel_size,
|
||
|
|
stride=self._stride,
|
||
|
|
dilation=self._dilation,
|
||
|
|
padding=self._padding,
|
||
|
|
)
|
||
|
|
|
||
|
|
self.set_h_init_to_uniform()
|
||
|
|
|
||
|
|
# ###############################################################
|
||
|
|
# Initialize the weights
|
||
|
|
# ###############################################################
|
||
|
|
|
||
|
|
assert len(self._weight_noise_range) == 2
|
||
|
|
weights = torch.empty(
|
||
|
|
(
|
||
|
|
int(self._kernel_size[0])
|
||
|
|
* int(self._kernel_size[1])
|
||
|
|
* int(self._number_of_input_neurons),
|
||
|
|
int(self._number_of_neurons),
|
||
|
|
),
|
||
|
|
dtype=self.default_dtype,
|
||
|
|
device=self.device,
|
||
|
|
)
|
||
|
|
|
||
|
|
torch.nn.init.uniform_(
|
||
|
|
weights,
|
||
|
|
a=float(self._weight_noise_range[0]),
|
||
|
|
b=float(self._weight_noise_range[1]),
|
||
|
|
)
|
||
|
|
self.weights = weights
|
||
|
|
|
||
|
|
self.functional_nnmf_sbs_bp = FunctionalNNMFSbSBP.apply
|
||
|
|
|
||
|
|
@property
|
||
|
|
def weights(self) -> torch.Tensor | None:
|
||
|
|
if self._weights_exists is False:
|
||
|
|
return None
|
||
|
|
else:
|
||
|
|
return self._weights
|
||
|
|
|
||
|
|
@weights.setter
|
||
|
|
def weights(self, value: torch.Tensor):
|
||
|
|
assert value is not None
|
||
|
|
assert torch.is_tensor(value) is True
|
||
|
|
assert value.dim() == 2
|
||
|
|
temp: torch.Tensor = (
|
||
|
|
value.detach()
|
||
|
|
.clone(memory_format=torch.contiguous_format)
|
||
|
|
.type(dtype=self.default_dtype)
|
||
|
|
.to(device=self.device)
|
||
|
|
)
|
||
|
|
temp /= temp.sum(dim=0, keepdim=True, dtype=self.default_dtype)
|
||
|
|
if self._weights_exists is False:
|
||
|
|
self._weights = torch.nn.parameter.Parameter(temp, requires_grad=True)
|
||
|
|
self._weights_exists = True
|
||
|
|
else:
|
||
|
|
self._weights.data = temp
|
||
|
|
|
||
|
|
@property
|
||
|
|
def h_initial(self) -> torch.Tensor | None:
|
||
|
|
return self._h_initial
|
||
|
|
|
||
|
|
@h_initial.setter
|
||
|
|
def h_initial(self, value: torch.Tensor):
|
||
|
|
assert value is not None
|
||
|
|
assert torch.is_tensor(value) is True
|
||
|
|
assert value.dim() == 1
|
||
|
|
assert value.dtype == self.default_dtype
|
||
|
|
self._h_initial = (
|
||
|
|
value.detach()
|
||
|
|
.clone(memory_format=torch.contiguous_format)
|
||
|
|
.type(dtype=self.default_dtype)
|
||
|
|
.to(device=self.device)
|
||
|
|
.requires_grad_(False)
|
||
|
|
)
|
||
|
|
|
||
|
|
def update_pre_care(self):
|
||
|
|
|
||
|
|
if self._weights.grad is not None:
|
||
|
|
assert self._number_of_grad_weight_contributions > 0
|
||
|
|
self._weights.grad /= self._number_of_grad_weight_contributions
|
||
|
|
self._number_of_grad_weight_contributions = 0.0
|
||
|
|
|
||
|
|
def update_after_care(self, threshold_weight: float):
|
||
|
|
|
||
|
|
if self._w_trainable is True:
|
||
|
|
self.norm_weights()
|
||
|
|
self.threshold_weights(threshold_weight)
|
||
|
|
self.norm_weights()
|
||
|
|
|
||
|
|
def after_batch(self, new_state: bool = False):
|
||
|
|
if self._keep_last_grad_scale is True:
|
||
|
|
self._last_grad_scale.data = self._last_grad_scale.grad
|
||
|
|
self._keep_last_grad_scale = new_state
|
||
|
|
|
||
|
|
self._last_grad_scale.grad = torch.zeros_like(self._last_grad_scale.grad)
|
||
|
|
|
||
|
|
def set_h_init_to_uniform(self) -> None:
|
||
|
|
|
||
|
|
assert self._number_of_neurons > 2
|
||
|
|
|
||
|
|
self.h_initial: torch.Tensor = torch.full(
|
||
|
|
(self._number_of_neurons,),
|
||
|
|
(1.0 / float(self._number_of_neurons)),
|
||
|
|
dtype=self.default_dtype,
|
||
|
|
device=self.device,
|
||
|
|
)
|
||
|
|
|
||
|
|
def norm_weights(self) -> None:
|
||
|
|
assert self._weights_exists is True
|
||
|
|
temp: torch.Tensor = (
|
||
|
|
self._weights.data.detach()
|
||
|
|
.clone(memory_format=torch.contiguous_format)
|
||
|
|
.type(dtype=self.default_dtype)
|
||
|
|
.to(device=self.device)
|
||
|
|
)
|
||
|
|
temp /= temp.sum(dim=0, keepdim=True, dtype=self.default_dtype)
|
||
|
|
self._weights.data = temp
|
||
|
|
|
||
|
|
def threshold_weights(self, threshold: float) -> None:
|
||
|
|
assert self._weights_exists is True
|
||
|
|
assert threshold >= 0
|
||
|
|
|
||
|
|
torch.clamp(
|
||
|
|
self._weights.data,
|
||
|
|
min=float(threshold),
|
||
|
|
max=None,
|
||
|
|
out=self._weights.data,
|
||
|
|
)
|
||
|
|
|
||
|
|
####################################################################
|
||
|
|
# Forward #
|
||
|
|
####################################################################
|
||
|
|
|
||
|
|
def forward(
|
||
|
|
self,
|
||
|
|
input: torch.Tensor,
|
||
|
|
) -> torch.Tensor:
|
||
|
|
|
||
|
|
# Are we happy with the input?
|
||
|
|
assert input is not None
|
||
|
|
assert torch.is_tensor(input) is True
|
||
|
|
assert input.dim() == 4
|
||
|
|
assert input.dtype == self.default_dtype
|
||
|
|
assert input.shape[1] == self._number_of_input_neurons
|
||
|
|
assert input.shape[2] == self._input_size[0]
|
||
|
|
assert input.shape[3] == self._input_size[1]
|
||
|
|
|
||
|
|
# Are we happy with the rest of the network?
|
||
|
|
assert self._epsilon_0 is not None
|
||
|
|
assert self._h_initial is not None
|
||
|
|
assert self._weights_exists is True
|
||
|
|
assert self._weights is not None
|
||
|
|
|
||
|
|
# Convolution of the input...
|
||
|
|
# Well, this is a convoltion layer
|
||
|
|
# there needs to be convolution somewhere
|
||
|
|
input_convolved = torch.nn.functional.fold(
|
||
|
|
torch.nn.functional.unfold(
|
||
|
|
input.requires_grad_(True),
|
||
|
|
kernel_size=(int(self._kernel_size[0]), int(self._kernel_size[1])),
|
||
|
|
dilation=(int(self._dilation[0]), int(self._dilation[1])),
|
||
|
|
padding=(int(self._padding[0]), int(self._padding[1])),
|
||
|
|
stride=(int(self._stride[0]), int(self._stride[1])),
|
||
|
|
),
|
||
|
|
output_size=tuple(self._output_size.tolist()),
|
||
|
|
kernel_size=(1, 1),
|
||
|
|
dilation=(1, 1),
|
||
|
|
padding=(0, 0),
|
||
|
|
stride=(1, 1),
|
||
|
|
)
|
||
|
|
|
||
|
|
# We might need the convolved input for other layers
|
||
|
|
# let us keep it for the future
|
||
|
|
if self.last_input_store is True:
|
||
|
|
self.last_input_data = input_convolved.detach().clone()
|
||
|
|
self.last_input_data /= self.last_input_data.sum(dim=1, keepdim=True)
|
||
|
|
else:
|
||
|
|
self.last_input_data = None
|
||
|
|
|
||
|
|
input_convolved = input_convolved / (
|
||
|
|
input_convolved.sum(dim=1, keepdim=True) + 1e-20
|
||
|
|
)
|
||
|
|
|
||
|
|
parameter_list = torch.tensor(
|
||
|
|
[
|
||
|
|
int(self._output_size[0]), # 0
|
||
|
|
int(self._output_size[1]), # 1
|
||
|
|
int(self._number_of_iterations), # 2
|
||
|
|
int(self._w_trainable), # 3
|
||
|
|
int(self._skip_gradient_calculation), # 4
|
||
|
|
int(self._keep_last_grad_scale), # 5
|
||
|
|
int(self._disable_scale_grade), # 6
|
||
|
|
],
|
||
|
|
dtype=torch.int64,
|
||
|
|
)
|
||
|
|
|
||
|
|
epsilon_0 = torch.tensor(self._epsilon_0)
|
||
|
|
|
||
|
|
h = self.functional_nnmf_sbs_bp(
|
||
|
|
input_convolved,
|
||
|
|
epsilon_0,
|
||
|
|
self._weights,
|
||
|
|
self._h_initial,
|
||
|
|
parameter_list,
|
||
|
|
self._last_grad_scale,
|
||
|
|
)
|
||
|
|
|
||
|
|
self._number_of_grad_weight_contributions += (
|
||
|
|
h.shape[0] * h.shape[-2] * h.shape[-1]
|
||
|
|
)
|
||
|
|
|
||
|
|
return h
|
||
|
|
|
||
|
|
|
||
|
|
class FunctionalNNMFSbSBP(torch.autograd.Function):
|
||
|
|
@staticmethod
|
||
|
|
def forward( # type: ignore
|
||
|
|
ctx,
|
||
|
|
input: torch.Tensor,
|
||
|
|
epsilon_0: torch.Tensor,
|
||
|
|
weights: torch.Tensor,
|
||
|
|
h_initial: torch.Tensor,
|
||
|
|
parameter_list: torch.Tensor,
|
||
|
|
grad_output_scale: torch.Tensor,
|
||
|
|
) -> torch.Tensor:
|
||
|
|
|
||
|
|
output_size_0: int = int(parameter_list[0])
|
||
|
|
output_size_1: int = int(parameter_list[1])
|
||
|
|
number_of_iterations: int = int(parameter_list[2])
|
||
|
|
|
||
|
|
_epsilon_0 = epsilon_0.item()
|
||
|
|
|
||
|
|
h = torch.tile(
|
||
|
|
h_initial.unsqueeze(0).unsqueeze(-1).unsqueeze(-1),
|
||
|
|
dims=[
|
||
|
|
int(input.shape[0]),
|
||
|
|
1,
|
||
|
|
int(output_size_0),
|
||
|
|
int(output_size_1),
|
||
|
|
],
|
||
|
|
)
|
||
|
|
|
||
|
|
for _ in range(0, number_of_iterations):
|
||
|
|
h_w = h.unsqueeze(1) * weights.unsqueeze(0).unsqueeze(-1).unsqueeze(-1)
|
||
|
|
h_w = h_w / (h_w.sum(dim=2, keepdim=True) + 1e-20)
|
||
|
|
h_w = (h_w * input.unsqueeze(2)).sum(dim=1)
|
||
|
|
if _epsilon_0 > 0:
|
||
|
|
h = h + _epsilon_0 * h_w
|
||
|
|
else:
|
||
|
|
h = h_w
|
||
|
|
h = h / (h.sum(dim=1, keepdim=True) + 1e-20)
|
||
|
|
|
||
|
|
ctx.save_for_backward(
|
||
|
|
input,
|
||
|
|
weights,
|
||
|
|
h,
|
||
|
|
parameter_list,
|
||
|
|
grad_output_scale,
|
||
|
|
)
|
||
|
|
|
||
|
|
return h
|
||
|
|
|
||
|
|
@staticmethod
|
||
|
|
def backward(ctx, grad_output):
|
||
|
|
# ##############################################
|
||
|
|
# Get the variables back
|
||
|
|
# ##############################################
|
||
|
|
(
|
||
|
|
input,
|
||
|
|
weights,
|
||
|
|
output,
|
||
|
|
parameter_list,
|
||
|
|
last_grad_scale,
|
||
|
|
) = ctx.saved_tensors
|
||
|
|
|
||
|
|
# ##############################################
|
||
|
|
# Default output
|
||
|
|
# ##############################################
|
||
|
|
grad_input = None
|
||
|
|
grad_epsilon_0 = None
|
||
|
|
grad_weights = None
|
||
|
|
grad_h_initial = None
|
||
|
|
grad_parameter_list = None
|
||
|
|
|
||
|
|
# ##############################################
|
||
|
|
# Parameters
|
||
|
|
# ##############################################
|
||
|
|
parameter_w_trainable: bool = bool(parameter_list[3])
|
||
|
|
parameter_skip_gradient_calculation: bool = bool(parameter_list[4])
|
||
|
|
parameter_keep_last_grad_scale: bool = bool(parameter_list[5])
|
||
|
|
parameter_disable_scale_grade: bool = bool(parameter_list[6])
|
||
|
|
|
||
|
|
# ##############################################
|
||
|
|
# Dealing with overall scale of the gradient
|
||
|
|
# ##############################################
|
||
|
|
if parameter_disable_scale_grade is False:
|
||
|
|
if parameter_keep_last_grad_scale is True:
|
||
|
|
last_grad_scale = torch.tensor(
|
||
|
|
[torch.abs(grad_output).max(), last_grad_scale]
|
||
|
|
).max()
|
||
|
|
grad_output /= last_grad_scale
|
||
|
|
grad_output_scale = last_grad_scale.clone()
|
||
|
|
|
||
|
|
input /= input.sum(dim=1, keepdim=True, dtype=weights.dtype) + 1e-20
|
||
|
|
|
||
|
|
# #################################################
|
||
|
|
# User doesn't want us to calculate the gradients
|
||
|
|
# #################################################
|
||
|
|
|
||
|
|
if parameter_skip_gradient_calculation is True:
|
||
|
|
|
||
|
|
return (
|
||
|
|
grad_input,
|
||
|
|
grad_epsilon_0,
|
||
|
|
grad_weights,
|
||
|
|
grad_h_initial,
|
||
|
|
grad_parameter_list,
|
||
|
|
grad_output_scale,
|
||
|
|
)
|
||
|
|
|
||
|
|
# #################################################
|
||
|
|
# Calculate backprop error (grad_input)
|
||
|
|
# #################################################
|
||
|
|
|
||
|
|
backprop_r: torch.Tensor = weights.unsqueeze(0).unsqueeze(-1).unsqueeze(
|
||
|
|
-1
|
||
|
|
) * output.unsqueeze(1)
|
||
|
|
|
||
|
|
backprop_bigr: torch.Tensor = backprop_r.sum(dim=2)
|
||
|
|
|
||
|
|
backprop_z: torch.Tensor = backprop_r * (
|
||
|
|
1.0 / (backprop_bigr + 1e-20)
|
||
|
|
).unsqueeze(2)
|
||
|
|
grad_input: torch.Tensor = (backprop_z * grad_output.unsqueeze(1)).sum(2)
|
||
|
|
del backprop_z
|
||
|
|
|
||
|
|
# #################################################
|
||
|
|
# Calculate weight gradient (grad_weights)
|
||
|
|
# #################################################
|
||
|
|
|
||
|
|
if parameter_w_trainable is False:
|
||
|
|
|
||
|
|
# #################################################
|
||
|
|
# We don't train this weight
|
||
|
|
# #################################################
|
||
|
|
grad_weights = None
|
||
|
|
|
||
|
|
else:
|
||
|
|
# #################################################
|
||
|
|
# Backprop
|
||
|
|
# #################################################
|
||
|
|
backprop_f: torch.Tensor = output.unsqueeze(1) * (
|
||
|
|
input / (backprop_bigr**2 + 1e-20)
|
||
|
|
).unsqueeze(2)
|
||
|
|
|
||
|
|
result_omega: torch.Tensor = backprop_bigr.unsqueeze(
|
||
|
|
2
|
||
|
|
) * grad_output.unsqueeze(1)
|
||
|
|
result_omega -= (backprop_r * grad_output.unsqueeze(1)).sum(2).unsqueeze(2)
|
||
|
|
result_omega *= backprop_f
|
||
|
|
del backprop_f
|
||
|
|
grad_weights = result_omega.sum(0).sum(-1).sum(-1)
|
||
|
|
del result_omega
|
||
|
|
|
||
|
|
del backprop_bigr
|
||
|
|
del backprop_r
|
||
|
|
|
||
|
|
return (
|
||
|
|
grad_input,
|
||
|
|
grad_epsilon_0,
|
||
|
|
grad_weights,
|
||
|
|
grad_h_initial,
|
||
|
|
grad_parameter_list,
|
||
|
|
grad_output_scale,
|
||
|
|
)
|