feat: working implementation of VAE
This commit is contained in:
@@ -121,7 +121,6 @@ class VariationalAutoencoder(AAutoencoder):
|
|||||||
def __init__(self,
|
def __init__(self,
|
||||||
encoder_layers: list[int],
|
encoder_layers: list[int],
|
||||||
decoder_layers: list[int],
|
decoder_layers: list[int],
|
||||||
sampling_size: int,
|
|
||||||
lr: float,
|
lr: float,
|
||||||
activation_func: ActivationFunc):
|
activation_func: ActivationFunc):
|
||||||
if encoder_layers[-1] != decoder_layers[0]:
|
if encoder_layers[-1] != decoder_layers[0]:
|
||||||
@@ -131,7 +130,12 @@ class VariationalAutoencoder(AAutoencoder):
|
|||||||
self.encoder = DeepNNLayer(encoder_layers, lr, activation_func)
|
self.encoder = DeepNNLayer(encoder_layers, lr, activation_func)
|
||||||
self.decoder = DeepNNLayer(decoder_layers, lr, activation_func)
|
self.decoder = DeepNNLayer(decoder_layers, lr, activation_func)
|
||||||
self.sampler = SampleLayer(self.encoder.out_size, lr, activation_func)
|
self.sampler = SampleLayer(self.encoder.out_size, lr, activation_func)
|
||||||
self.sampling_size = sampling_size
|
|
||||||
|
def loss(self, data_set: list[np.ndarray]) -> float:
|
||||||
|
loss = 0
|
||||||
|
for x in data_set:
|
||||||
|
loss += np.sum(np.abs(x - self.forward(x)[0])) / len(x)
|
||||||
|
return loss / len(data_set)
|
||||||
|
|
||||||
def load(path: str) -> 'ClassicalAutoencoder':
|
def load(path: str) -> 'ClassicalAutoencoder':
|
||||||
path = path.removesuffix('.npy') + '.npy'
|
path = path.removesuffix('.npy') + '.npy'
|
||||||
@@ -139,16 +143,17 @@ class VariationalAutoencoder(AAutoencoder):
|
|||||||
return data.item()
|
return data.item()
|
||||||
|
|
||||||
def train(self, v: np.ndarray) -> float:
|
def train(self, v: np.ndarray) -> float:
|
||||||
out_enc = self.encoder.forward(v)
|
out = self.forward(v)
|
||||||
in_samples = np.zeros(
|
error = out - v
|
||||||
(self.sampling_size, self.encoder.out_size)
|
self.encoder.backprop(
|
||||||
|
self.sampler.backprop(
|
||||||
|
self.decoder.backprop(error)
|
||||||
)
|
)
|
||||||
out_samples = np.zeros(
|
|
||||||
(self.sampling_size, self.decoder.out_size)
|
|
||||||
)
|
)
|
||||||
for i in range(self.sampling_size):
|
return np.sum(np.abs(error)) / len(v)
|
||||||
in_samples[i] = self.sampler.forward(out_enc)
|
|
||||||
out_samples[i] = self.decoder.forward(in_samples[i])
|
|
||||||
|
|
||||||
def forward(self, v: np.ndarray) -> np.ndarray:
|
def forward(self, v: np.ndarray) -> np.ndarray:
|
||||||
pass
|
code = self.encoder.forward(v)
|
||||||
|
sample = self.sampler.forward(code)
|
||||||
|
out = self.decoder.forward(sample)
|
||||||
|
return out, code
|
||||||
|
|||||||
15
layers.py
15
layers.py
@@ -29,7 +29,7 @@ class NNLayer:
|
|||||||
return self.output
|
return self.output
|
||||||
|
|
||||||
def backprop(self, error: np.ndarray) -> np.ndarray:
|
def backprop(self, error: np.ndarray) -> np.ndarray:
|
||||||
error *= self.activation_func.derivative(self.output_linear)
|
error *= self.activation_func.d(self.output_linear)
|
||||||
ret = self.W @ error
|
ret = self.W @ error
|
||||||
dW = np.outer(self.input, error) * self.lr
|
dW = np.outer(self.input, error) * self.lr
|
||||||
dB = error * self.lr
|
dB = error * self.lr
|
||||||
@@ -57,12 +57,15 @@ class SampleLayer:
|
|||||||
|
|
||||||
def forward(self, v: np.ndarray) -> np.ndarray:
|
def forward(self, v: np.ndarray) -> np.ndarray:
|
||||||
self.input = v
|
self.input = v
|
||||||
mean = self.mean_nn.forward(v)
|
self.mean = self.mean_nn.forward(v)
|
||||||
std = self.std_nn.forward(v)
|
self.std = self.std_nn.forward(v)
|
||||||
return np.random.normal(mean, std, 1)
|
self.eps = np.random.normal(0, 1)
|
||||||
|
return self.eps * self.std + self.mean
|
||||||
|
|
||||||
def backprop(self, errors: np.ndarray) -> np.ndarray:
|
def backprop(self, error: np.ndarray) -> np.ndarray:
|
||||||
pass
|
mu_error = self.mean_nn.backprop(error)
|
||||||
|
std_error = self.std_nn.backprop(self.eps * error)
|
||||||
|
return mu_error + std_error
|
||||||
|
|
||||||
|
|
||||||
class DeepNNLayer:
|
class DeepNNLayer:
|
||||||
|
|||||||
@@ -1,6 +1,6 @@
|
|||||||
import matplotlib.pyplot as plt
|
import matplotlib.pyplot as plt
|
||||||
import numpy as np
|
import numpy as np
|
||||||
from autoencoder import ClassicalAutoencoder
|
from autoencoder import VariationalAutoencoder, AAutoencoder
|
||||||
from activations import LeakyReLU
|
from activations import LeakyReLU
|
||||||
import os
|
import os
|
||||||
|
|
||||||
@@ -21,19 +21,21 @@ def mnist_train(
|
|||||||
filename: str,
|
filename: str,
|
||||||
max_epoch: int,
|
max_epoch: int,
|
||||||
patience: int,
|
patience: int,
|
||||||
) -> ClassicalAutoencoder:
|
cls: type[AAutoencoder]
|
||||||
|
) -> AAutoencoder:
|
||||||
x_train, _, x_test, _ = load_mnist()
|
x_train, _, x_test, _ = load_mnist()
|
||||||
in_len = x_train[0].shape[0] * x_train[0].shape[0]
|
in_len = x_train[0].shape[0] * x_train[0].shape[0]
|
||||||
x_train.resize(x_train.shape[0], in_len)
|
x_train.resize(x_train.shape[0], in_len)
|
||||||
x_test.resize(x_test.shape[0], in_len)
|
x_test.resize(x_test.shape[0], in_len)
|
||||||
x_train = x_train / 255
|
x_train = x_train / 255
|
||||||
x_test = x_test / 255
|
x_test = x_test / 255
|
||||||
|
x_train = x_train[:5000]
|
||||||
if os.path.exists(filename):
|
if os.path.exists(filename):
|
||||||
autoencoder = ClassicalAutoencoder.load(filename)
|
autoencoder = cls.load(filename)
|
||||||
else:
|
else:
|
||||||
autoencoder = ClassicalAutoencoder(
|
autoencoder = cls(
|
||||||
[in_len, 64, 16],
|
[in_len, 16],
|
||||||
[16, 64, in_len],
|
[16, in_len],
|
||||||
0.01,
|
0.01,
|
||||||
LeakyReLU()
|
LeakyReLU()
|
||||||
)
|
)
|
||||||
@@ -46,7 +48,7 @@ def mnist_train(
|
|||||||
return autoencoder
|
return autoencoder
|
||||||
|
|
||||||
|
|
||||||
def mnist_test(model: str | ClassicalAutoencoder):
|
def mnist_test(model: str | AAutoencoder):
|
||||||
x_train, _, x_test, y_test = load_mnist()
|
x_train, _, x_test, y_test = load_mnist()
|
||||||
in_len = x_train[0].shape[0] * x_train[0].shape[0]
|
in_len = x_train[0].shape[0] * x_train[0].shape[0]
|
||||||
img_shape = x_train[0].shape
|
img_shape = x_train[0].shape
|
||||||
@@ -55,7 +57,7 @@ def mnist_test(model: str | ClassicalAutoencoder):
|
|||||||
x_train = x_train / 255
|
x_train = x_train / 255
|
||||||
x_test = x_test / 255
|
x_test = x_test / 255
|
||||||
if isinstance(model, str):
|
if isinstance(model, str):
|
||||||
autoencoder: ClassicalAutoencoder = ClassicalAutoencoder.load(model)
|
autoencoder: AAutoencoder = AAutoencoder.load(model)
|
||||||
else:
|
else:
|
||||||
autoencoder = model
|
autoencoder = model
|
||||||
print(autoencoder)
|
print(autoencoder)
|
||||||
@@ -114,5 +116,10 @@ if __name__ == "__main__":
|
|||||||
if args.r:
|
if args.r:
|
||||||
mnist_test(args.m)
|
mnist_test(args.m)
|
||||||
else:
|
else:
|
||||||
autoencoder = mnist_train(args.m, args.e, args.p)
|
autoencoder = mnist_train(
|
||||||
|
args.m,
|
||||||
|
args.e,
|
||||||
|
args.p,
|
||||||
|
VariationalAutoencoder
|
||||||
|
)
|
||||||
mnist_test(autoencoder)
|
mnist_test(autoencoder)
|
||||||
|
|||||||
Reference in New Issue
Block a user