Skip to content

random_gaussian_noise

Random Gaussian Noise Layer API

This module provides classes to build random Gaussian noise layers.

Classes:

Classes

RandomGaussianNoise1D 

RandomGaussianNoise1D(factor: float | tuple[float, float] = 0.1, **kwargs)

Apply additive zero-centered Gaussian noise.

Parameters:

  • factor (float, default: 0.1 ) –

    Standard deviation of the Gaussian noise.

Example:

    x = np.sin(2*np.pi*10*np.arange(duration_size)/100)
    lyr = RandomGaussianNoise1D(factor=0.1)
    y = lyr(x)
Source code in neuralspot_edge/layers/preprocessing/random_gaussian_noise.py 
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
def __init__(self, factor: float | tuple[float, float] = 0.1, **kwargs):
    """Apply additive zero-centered Gaussian noise.

    Args:
        factor (float): Standard deviation of the Gaussian noise.

    Example:

    ```python
        x = np.sin(2*np.pi*10*np.arange(duration_size)/100)
        lyr = RandomGaussianNoise1D(factor=0.1)
        y = lyr(x)
    ```
    """
    super().__init__(**kwargs)

    self.factor = parse_factor(factor, min_value=0, max_value=None, param_name="factor")

Functions

get_random_transformations 
get_random_transformations(input_shape: tuple[int, ...]) -> dict

Generate noise tensor

Parameters:

  • input_shape (tuple[int, ...]) –

    Input shape.

Returns:

  • dict ( dict ) –

    Dictionary containing the noise tensor.

Source code in neuralspot_edge/layers/preprocessing/random_gaussian_noise.py 
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
def get_random_transformations(self, input_shape: tuple[int, ...]) -> dict:
    """Generate noise tensor

    Args:
        input_shape (tuple[int, ...]): Input shape.

    Returns:
        dict: Dictionary containing the noise tensor.
    """
    stddev = keras.random.uniform(
        shape=(),
        minval=self.factor[0],
        maxval=self.factor[1],
        seed=self.random_generator,
        dtype=self.compute_dtype,
    )
    return {
        "noise": keras.random.normal(
            shape=input_shape, stddev=stddev, dtype=self.compute_dtype, seed=self.random_generator
        )
    }
augment_samples 
augment_samples(inputs) -> keras.KerasTensor

Augment all samples in the batch as it's faster.

Source code in neuralspot_edge/layers/preprocessing/random_gaussian_noise.py 
60
61
62
63
64
65
66
def augment_samples(self, inputs) -> keras.KerasTensor:
    """Augment all samples in the batch as it's faster."""
    samples = inputs[self.SAMPLES]
    if self.training:
        noise = inputs[self.TRANSFORMS]["noise"]
        return samples + noise
    return samples