def __init__(self,
x,
filters,
filter_size,
step,
bias=True,
activation=None,
padding='VALID',
*activation_args,
**activation_kwargs):
if isinstance(x, Layer):
x = x.output
self.kernel = graph.Variable(
(filter_size, filter_size, x.shape[2], filters))
y = graph.Convolution(x, self.kernel, step, padding)
self.bias = graph.Variable((filters, )) if bias else None
if bias:
y = graph.Sum(y, self.bias)
if activation:
y = activation(y, *activation_args, **activation_kwargs)
super(Convolution, self).__init__((x, ), y, [self.kernel, self.bias])
def test_convolution_with_step(self):
x = graph.Constant(self.x)
f = graph.Constant(self.filters)
conv = graph.Convolution(x, f, step=2)
np.testing.assert_array_equal(self.expected_step_2, graph.run(conv))
def test_convolution(self):
x_c = graph.Constant(self.x)
f_c = graph.Constant(self.filters)
conv = graph.Convolution(x_c, f_c)
np.testing.assert_array_equal(graph.run(conv), self.expected)
def test_convolution_batched(self):
x_p = graph.Placeholder(self.x.shape, batched=True)
f_c = graph.Constant(self.filters)
conv = graph.Convolution(x_p, f_c)
x = np.stack([self.x, 2 * self.x, 3 * self.x])
expected = np.stack(
[self.expected, 2 * self.expected, 3 * self.expected])
np.testing.assert_array_equal(graph.run(conv, {x_p: x}), expected)
def test_convolution_grad_with_step(self):
x_c = graph.Constant(self.x)
f_c = graph.Constant(self.filters)
conv = graph.Convolution(x_c, f_c, step=2)
grad = graph.Grad(conv, [x_c, f_c])
grad_x, grad_f = graph.run(grad)
np.testing.assert_array_equal(grad_x, self.grad_x_step_2)
np.testing.assert_array_equal(grad_f, self.grad_f_step_2)
def test_convolution_grad_batched(self):
x_p = graph.Placeholder(self.x.shape, batched=True)
f_c = graph.Constant(self.filters)
x = np.stack([self.x, 2 * self.x, 3 * self.x])
conv = graph.Convolution(x_p, f_c)
grad = graph.Grad(conv, [x_p, f_c])
grad_x, grad_f = graph.run(grad, {x_p: x})
np.testing.assert_array_equal(grad_x, np.stack([self.grad_x] * 3))
np.testing.assert_array_equal(grad_f, self.grad_f * 6)